{ "best_metric": null, "best_model_checkpoint": null, "epoch": 5.0, "eval_steps": 2500, "global_step": 33755, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 2.902843601895735e-07, "loss": 9.1546, "step": 50 }, { "epoch": 0.01, "learning_rate": 5.80568720379147e-07, "loss": 9.1497, "step": 100 }, { "epoch": 0.02, "learning_rate": 8.767772511848342e-07, "loss": 9.065, "step": 150 }, { "epoch": 0.03, "learning_rate": 1.1729857819905214e-06, "loss": 8.9258, "step": 200 }, { "epoch": 0.04, "learning_rate": 1.4632701421800948e-06, "loss": 8.8689, "step": 250 }, { "epoch": 0.04, "learning_rate": 1.7594786729857822e-06, "loss": 8.4535, "step": 300 }, { "epoch": 0.05, "learning_rate": 2.055687203791469e-06, "loss": 7.775, "step": 350 }, { "epoch": 0.06, "learning_rate": 2.3518957345971564e-06, "loss": 7.0715, "step": 400 }, { "epoch": 0.07, "learning_rate": 2.648104265402844e-06, "loss": 6.3666, "step": 450 }, { "epoch": 0.07, "learning_rate": 2.944312796208531e-06, "loss": 5.8049, "step": 500 }, { "epoch": 0.08, "learning_rate": 3.2405213270142184e-06, "loss": 5.1548, "step": 550 }, { "epoch": 0.09, "learning_rate": 3.5367298578199056e-06, "loss": 4.8392, "step": 600 }, { "epoch": 0.1, "learning_rate": 3.832938388625592e-06, "loss": 4.3625, "step": 650 }, { "epoch": 0.1, "learning_rate": 4.12914691943128e-06, "loss": 4.0536, "step": 700 }, { "epoch": 0.11, "learning_rate": 4.425355450236968e-06, "loss": 3.8893, "step": 750 }, { "epoch": 0.12, "learning_rate": 4.721563981042654e-06, "loss": 3.673, "step": 800 }, { "epoch": 0.13, "learning_rate": 5.017772511848341e-06, "loss": 3.5412, "step": 850 }, { "epoch": 0.13, "learning_rate": 5.313981042654029e-06, "loss": 3.3889, "step": 900 }, { "epoch": 0.14, "learning_rate": 5.6101895734597155e-06, "loss": 3.2857, "step": 950 }, { "epoch": 0.15, "learning_rate": 5.906398104265403e-06, "loss": 3.257, "step": 1000 }, { "epoch": 0.16, "learning_rate": 6.20260663507109e-06, "loss": 3.1638, "step": 1050 }, { "epoch": 0.16, "learning_rate": 6.4988151658767776e-06, "loss": 3.123, "step": 1100 }, { "epoch": 0.17, "learning_rate": 6.795023696682464e-06, "loss": 3.0633, "step": 1150 }, { "epoch": 0.18, "learning_rate": 7.091232227488152e-06, "loss": 3.0384, "step": 1200 }, { "epoch": 0.19, "learning_rate": 7.387440758293839e-06, "loss": 3.0122, "step": 1250 }, { "epoch": 0.19, "learning_rate": 7.683649289099525e-06, "loss": 2.9837, "step": 1300 }, { "epoch": 0.2, "learning_rate": 7.979857819905213e-06, "loss": 2.9659, "step": 1350 }, { "epoch": 0.21, "learning_rate": 8.2760663507109e-06, "loss": 2.9442, "step": 1400 }, { "epoch": 0.21, "learning_rate": 8.572274881516588e-06, "loss": 2.9377, "step": 1450 }, { "epoch": 0.22, "learning_rate": 8.868483412322276e-06, "loss": 2.9212, "step": 1500 }, { "epoch": 0.23, "learning_rate": 9.164691943127962e-06, "loss": 2.9104, "step": 1550 }, { "epoch": 0.24, "learning_rate": 9.46090047393365e-06, "loss": 2.9005, "step": 1600 }, { "epoch": 0.24, "learning_rate": 9.757109004739337e-06, "loss": 2.8978, "step": 1650 }, { "epoch": 0.25, "learning_rate": 1.0053317535545025e-05, "loss": 2.8991, "step": 1700 }, { "epoch": 0.26, "learning_rate": 1.0349526066350712e-05, "loss": 2.8838, "step": 1750 }, { "epoch": 0.27, "learning_rate": 1.0645734597156398e-05, "loss": 2.8798, "step": 1800 }, { "epoch": 0.27, "learning_rate": 1.0941943127962088e-05, "loss": 2.8755, "step": 1850 }, { "epoch": 0.28, "learning_rate": 1.1238151658767774e-05, "loss": 2.8707, "step": 1900 }, { "epoch": 0.29, "learning_rate": 1.1534360189573461e-05, "loss": 2.8754, "step": 1950 }, { "epoch": 0.3, "learning_rate": 1.1830568720379147e-05, "loss": 2.8679, "step": 2000 }, { "epoch": 0.3, "learning_rate": 1.2126777251184836e-05, "loss": 2.8694, "step": 2050 }, { "epoch": 0.31, "learning_rate": 1.2422985781990522e-05, "loss": 2.8627, "step": 2100 }, { "epoch": 0.32, "learning_rate": 1.271919431279621e-05, "loss": 2.8631, "step": 2150 }, { "epoch": 0.33, "learning_rate": 1.3015402843601896e-05, "loss": 2.8616, "step": 2200 }, { "epoch": 0.33, "learning_rate": 1.3311611374407585e-05, "loss": 2.8606, "step": 2250 }, { "epoch": 0.34, "learning_rate": 1.3607819905213271e-05, "loss": 2.8607, "step": 2300 }, { "epoch": 0.35, "learning_rate": 1.3904028436018959e-05, "loss": 2.8573, "step": 2350 }, { "epoch": 0.36, "learning_rate": 1.4200236966824645e-05, "loss": 2.8597, "step": 2400 }, { "epoch": 0.36, "learning_rate": 1.4496445497630334e-05, "loss": 2.8555, "step": 2450 }, { "epoch": 0.37, "learning_rate": 1.479265402843602e-05, "loss": 2.8545, "step": 2500 }, { "epoch": 0.37, "eval_loss": 2.887249708175659, "eval_runtime": 348.4295, "eval_samples_per_second": 15.954, "eval_steps_per_second": 1.995, "eval_wer": 1.0, "step": 2500 }, { "epoch": 0.38, "learning_rate": 1.5088862559241708e-05, "loss": 2.8542, "step": 2550 }, { "epoch": 0.39, "learning_rate": 1.5385071090047395e-05, "loss": 2.8544, "step": 2600 }, { "epoch": 0.39, "learning_rate": 1.568127962085308e-05, "loss": 2.8561, "step": 2650 }, { "epoch": 0.4, "learning_rate": 1.597748815165877e-05, "loss": 2.8503, "step": 2700 }, { "epoch": 0.41, "learning_rate": 1.6273696682464456e-05, "loss": 2.853, "step": 2750 }, { "epoch": 0.41, "learning_rate": 1.6569905213270142e-05, "loss": 2.8523, "step": 2800 }, { "epoch": 0.42, "learning_rate": 1.686611374407583e-05, "loss": 2.8522, "step": 2850 }, { "epoch": 0.43, "learning_rate": 1.7162322274881517e-05, "loss": 2.8529, "step": 2900 }, { "epoch": 0.44, "learning_rate": 1.7458530805687207e-05, "loss": 2.8445, "step": 2950 }, { "epoch": 0.44, "learning_rate": 1.7754739336492893e-05, "loss": 2.8419, "step": 3000 }, { "epoch": 0.45, "learning_rate": 1.8050947867298582e-05, "loss": 2.8349, "step": 3050 }, { "epoch": 0.46, "learning_rate": 1.8347156398104268e-05, "loss": 2.8246, "step": 3100 }, { "epoch": 0.47, "learning_rate": 1.8643364928909954e-05, "loss": 2.7982, "step": 3150 }, { "epoch": 0.47, "learning_rate": 1.893957345971564e-05, "loss": 2.7596, "step": 3200 }, { "epoch": 0.48, "learning_rate": 1.923578199052133e-05, "loss": 2.72, "step": 3250 }, { "epoch": 0.49, "learning_rate": 1.9531990521327015e-05, "loss": 2.6601, "step": 3300 }, { "epoch": 0.5, "learning_rate": 1.9828199052132704e-05, "loss": 2.5809, "step": 3350 }, { "epoch": 0.5, "learning_rate": 1.998617466012706e-05, "loss": 2.5155, "step": 3400 }, { "epoch": 0.51, "learning_rate": 1.9953257184239115e-05, "loss": 2.4071, "step": 3450 }, { "epoch": 0.52, "learning_rate": 1.9920339708351166e-05, "loss": 2.2622, "step": 3500 }, { "epoch": 0.53, "learning_rate": 1.9887422232463217e-05, "loss": 2.1511, "step": 3550 }, { "epoch": 0.53, "learning_rate": 1.985450475657527e-05, "loss": 2.0263, "step": 3600 }, { "epoch": 0.54, "learning_rate": 1.9821587280687316e-05, "loss": 1.9196, "step": 3650 }, { "epoch": 0.55, "learning_rate": 1.978866980479937e-05, "loss": 1.7882, "step": 3700 }, { "epoch": 0.56, "learning_rate": 1.975575232891142e-05, "loss": 1.6822, "step": 3750 }, { "epoch": 0.56, "learning_rate": 1.9722834853023473e-05, "loss": 1.5541, "step": 3800 }, { "epoch": 0.57, "learning_rate": 1.9689917377135524e-05, "loss": 1.5421, "step": 3850 }, { "epoch": 0.58, "learning_rate": 1.9656999901247575e-05, "loss": 1.4162, "step": 3900 }, { "epoch": 0.59, "learning_rate": 1.9624082425359626e-05, "loss": 1.323, "step": 3950 }, { "epoch": 0.59, "learning_rate": 1.9591164949471677e-05, "loss": 1.2838, "step": 4000 }, { "epoch": 0.6, "learning_rate": 1.9558247473583728e-05, "loss": 1.1865, "step": 4050 }, { "epoch": 0.61, "learning_rate": 1.952532999769578e-05, "loss": 1.1166, "step": 4100 }, { "epoch": 0.61, "learning_rate": 1.949241252180783e-05, "loss": 1.1203, "step": 4150 }, { "epoch": 0.62, "learning_rate": 1.945949504591988e-05, "loss": 1.0965, "step": 4200 }, { "epoch": 0.63, "learning_rate": 1.942657757003193e-05, "loss": 0.9832, "step": 4250 }, { "epoch": 0.64, "learning_rate": 1.9393660094143983e-05, "loss": 0.9714, "step": 4300 }, { "epoch": 0.64, "learning_rate": 1.9360742618256034e-05, "loss": 0.9182, "step": 4350 }, { "epoch": 0.65, "learning_rate": 1.9327825142368085e-05, "loss": 0.9218, "step": 4400 }, { "epoch": 0.66, "learning_rate": 1.9294907666480136e-05, "loss": 0.9231, "step": 4450 }, { "epoch": 0.67, "learning_rate": 1.9261990190592187e-05, "loss": 0.8684, "step": 4500 }, { "epoch": 0.67, "learning_rate": 1.9229072714704238e-05, "loss": 0.8196, "step": 4550 }, { "epoch": 0.68, "learning_rate": 1.919615523881629e-05, "loss": 0.8499, "step": 4600 }, { "epoch": 0.69, "learning_rate": 1.916323776292834e-05, "loss": 0.8108, "step": 4650 }, { "epoch": 0.7, "learning_rate": 1.913032028704039e-05, "loss": 0.7843, "step": 4700 }, { "epoch": 0.7, "learning_rate": 1.9097402811152442e-05, "loss": 0.7981, "step": 4750 }, { "epoch": 0.71, "learning_rate": 1.9064485335264493e-05, "loss": 0.7909, "step": 4800 }, { "epoch": 0.72, "learning_rate": 1.9031567859376544e-05, "loss": 0.7549, "step": 4850 }, { "epoch": 0.73, "learning_rate": 1.8998650383488595e-05, "loss": 0.7125, "step": 4900 }, { "epoch": 0.73, "learning_rate": 1.8965732907600646e-05, "loss": 0.7799, "step": 4950 }, { "epoch": 0.74, "learning_rate": 1.89328154317127e-05, "loss": 0.7012, "step": 5000 }, { "epoch": 0.74, "eval_loss": 0.34729066491127014, "eval_runtime": 94.3417, "eval_samples_per_second": 58.924, "eval_steps_per_second": 7.367, "eval_wer": 0.28398097580037934, "step": 5000 }, { "epoch": 0.75, "learning_rate": 1.8899897955824748e-05, "loss": 0.7523, "step": 5050 }, { "epoch": 0.76, "learning_rate": 1.88669804799368e-05, "loss": 0.6433, "step": 5100 }, { "epoch": 0.76, "learning_rate": 1.883406300404885e-05, "loss": 0.635, "step": 5150 }, { "epoch": 0.77, "learning_rate": 1.88011455281609e-05, "loss": 0.6563, "step": 5200 }, { "epoch": 0.78, "learning_rate": 1.8768228052272955e-05, "loss": 0.6238, "step": 5250 }, { "epoch": 0.79, "learning_rate": 1.8735310576385003e-05, "loss": 0.595, "step": 5300 }, { "epoch": 0.79, "learning_rate": 1.8702393100497057e-05, "loss": 0.6241, "step": 5350 }, { "epoch": 0.8, "learning_rate": 1.8669475624609105e-05, "loss": 0.6602, "step": 5400 }, { "epoch": 0.81, "learning_rate": 1.863655814872116e-05, "loss": 0.6206, "step": 5450 }, { "epoch": 0.81, "learning_rate": 1.860364067283321e-05, "loss": 0.6292, "step": 5500 }, { "epoch": 0.82, "learning_rate": 1.8570723196945258e-05, "loss": 0.6009, "step": 5550 }, { "epoch": 0.83, "learning_rate": 1.8537805721057312e-05, "loss": 0.6348, "step": 5600 }, { "epoch": 0.84, "learning_rate": 1.850488824516936e-05, "loss": 0.572, "step": 5650 }, { "epoch": 0.84, "learning_rate": 1.8471970769281414e-05, "loss": 0.5527, "step": 5700 }, { "epoch": 0.85, "learning_rate": 1.8439053293393466e-05, "loss": 0.5879, "step": 5750 }, { "epoch": 0.86, "learning_rate": 1.8406135817505517e-05, "loss": 0.5748, "step": 5800 }, { "epoch": 0.87, "learning_rate": 1.8373218341617568e-05, "loss": 0.5542, "step": 5850 }, { "epoch": 0.87, "learning_rate": 1.8340300865729615e-05, "loss": 0.5357, "step": 5900 }, { "epoch": 0.88, "learning_rate": 1.830738338984167e-05, "loss": 0.5334, "step": 5950 }, { "epoch": 0.89, "learning_rate": 1.827446591395372e-05, "loss": 0.5453, "step": 6000 }, { "epoch": 0.9, "learning_rate": 1.824154843806577e-05, "loss": 0.5886, "step": 6050 }, { "epoch": 0.9, "learning_rate": 1.8208630962177823e-05, "loss": 0.5514, "step": 6100 }, { "epoch": 0.91, "learning_rate": 1.8175713486289874e-05, "loss": 0.5222, "step": 6150 }, { "epoch": 0.92, "learning_rate": 1.8142796010401925e-05, "loss": 0.5379, "step": 6200 }, { "epoch": 0.93, "learning_rate": 1.8109878534513976e-05, "loss": 0.5194, "step": 6250 }, { "epoch": 0.93, "learning_rate": 1.8076961058626027e-05, "loss": 0.5172, "step": 6300 }, { "epoch": 0.94, "learning_rate": 1.8044043582738078e-05, "loss": 0.5348, "step": 6350 }, { "epoch": 0.95, "learning_rate": 1.801112610685013e-05, "loss": 0.4841, "step": 6400 }, { "epoch": 0.96, "learning_rate": 1.797820863096218e-05, "loss": 0.5201, "step": 6450 }, { "epoch": 0.96, "learning_rate": 1.794529115507423e-05, "loss": 0.5244, "step": 6500 }, { "epoch": 0.97, "learning_rate": 1.7912373679186282e-05, "loss": 0.4829, "step": 6550 }, { "epoch": 0.98, "learning_rate": 1.7880114552816094e-05, "loss": 0.521, "step": 6600 }, { "epoch": 0.99, "learning_rate": 1.784719707692814e-05, "loss": 0.4329, "step": 6650 }, { "epoch": 0.99, "learning_rate": 1.7814279601040196e-05, "loss": 0.5034, "step": 6700 }, { "epoch": 1.0, "learning_rate": 1.7781362125152243e-05, "loss": 0.4883, "step": 6750 }, { "epoch": 1.01, "learning_rate": 1.7748444649264294e-05, "loss": 0.5063, "step": 6800 }, { "epoch": 1.01, "learning_rate": 1.771552717337635e-05, "loss": 0.5545, "step": 6850 }, { "epoch": 1.02, "learning_rate": 1.7682609697488396e-05, "loss": 0.4859, "step": 6900 }, { "epoch": 1.03, "learning_rate": 1.764969222160045e-05, "loss": 0.4756, "step": 6950 }, { "epoch": 1.04, "learning_rate": 1.76167747457125e-05, "loss": 0.4688, "step": 7000 }, { "epoch": 1.04, "learning_rate": 1.7583857269824553e-05, "loss": 0.5528, "step": 7050 }, { "epoch": 1.05, "learning_rate": 1.7550939793936604e-05, "loss": 0.462, "step": 7100 }, { "epoch": 1.06, "learning_rate": 1.751802231804865e-05, "loss": 0.4634, "step": 7150 }, { "epoch": 1.07, "learning_rate": 1.7485104842160706e-05, "loss": 0.5026, "step": 7200 }, { "epoch": 1.07, "learning_rate": 1.7452187366272754e-05, "loss": 0.4976, "step": 7250 }, { "epoch": 1.08, "learning_rate": 1.7419269890384808e-05, "loss": 0.4767, "step": 7300 }, { "epoch": 1.09, "learning_rate": 1.738635241449686e-05, "loss": 0.4584, "step": 7350 }, { "epoch": 1.1, "learning_rate": 1.735343493860891e-05, "loss": 0.4556, "step": 7400 }, { "epoch": 1.1, "learning_rate": 1.732051746272096e-05, "loss": 0.4692, "step": 7450 }, { "epoch": 1.11, "learning_rate": 1.728759998683301e-05, "loss": 0.46, "step": 7500 }, { "epoch": 1.11, "eval_loss": 0.20323950052261353, "eval_runtime": 93.8223, "eval_samples_per_second": 59.25, "eval_steps_per_second": 7.408, "eval_wer": 0.1510022016984531, "step": 7500 }, { "epoch": 1.12, "learning_rate": 1.7254682510945063e-05, "loss": 0.4607, "step": 7550 }, { "epoch": 1.13, "learning_rate": 1.7221765035057114e-05, "loss": 0.4231, "step": 7600 }, { "epoch": 1.13, "learning_rate": 1.7188847559169165e-05, "loss": 0.4986, "step": 7650 }, { "epoch": 1.14, "learning_rate": 1.7155930083281216e-05, "loss": 0.4645, "step": 7700 }, { "epoch": 1.15, "learning_rate": 1.7123012607393267e-05, "loss": 0.4507, "step": 7750 }, { "epoch": 1.16, "learning_rate": 1.7090095131505318e-05, "loss": 0.4512, "step": 7800 }, { "epoch": 1.16, "learning_rate": 1.705717765561737e-05, "loss": 0.4333, "step": 7850 }, { "epoch": 1.17, "learning_rate": 1.702426017972942e-05, "loss": 0.4414, "step": 7900 }, { "epoch": 1.18, "learning_rate": 1.699134270384147e-05, "loss": 0.449, "step": 7950 }, { "epoch": 1.19, "learning_rate": 1.6958425227953522e-05, "loss": 0.4408, "step": 8000 }, { "epoch": 1.19, "learning_rate": 1.6925507752065573e-05, "loss": 0.4323, "step": 8050 }, { "epoch": 1.2, "learning_rate": 1.6892590276177624e-05, "loss": 0.4439, "step": 8100 }, { "epoch": 1.21, "learning_rate": 1.6859672800289675e-05, "loss": 0.4241, "step": 8150 }, { "epoch": 1.21, "learning_rate": 1.6826755324401726e-05, "loss": 0.4587, "step": 8200 }, { "epoch": 1.22, "learning_rate": 1.6793837848513777e-05, "loss": 0.4262, "step": 8250 }, { "epoch": 1.23, "learning_rate": 1.6760920372625828e-05, "loss": 0.4298, "step": 8300 }, { "epoch": 1.24, "learning_rate": 1.672800289673788e-05, "loss": 0.4084, "step": 8350 }, { "epoch": 1.24, "learning_rate": 1.669508542084993e-05, "loss": 0.4762, "step": 8400 }, { "epoch": 1.25, "learning_rate": 1.666216794496198e-05, "loss": 0.411, "step": 8450 }, { "epoch": 1.26, "learning_rate": 1.6629250469074032e-05, "loss": 0.4639, "step": 8500 }, { "epoch": 1.27, "learning_rate": 1.6596332993186083e-05, "loss": 0.3964, "step": 8550 }, { "epoch": 1.27, "learning_rate": 1.6563415517298134e-05, "loss": 0.424, "step": 8600 }, { "epoch": 1.28, "learning_rate": 1.6530498041410185e-05, "loss": 0.4324, "step": 8650 }, { "epoch": 1.29, "learning_rate": 1.6497580565522236e-05, "loss": 0.4041, "step": 8700 }, { "epoch": 1.3, "learning_rate": 1.6464663089634287e-05, "loss": 0.4354, "step": 8750 }, { "epoch": 1.3, "learning_rate": 1.643174561374634e-05, "loss": 0.3862, "step": 8800 }, { "epoch": 1.31, "learning_rate": 1.6399486487376147e-05, "loss": 0.3986, "step": 8850 }, { "epoch": 1.32, "learning_rate": 1.63665690114882e-05, "loss": 0.3746, "step": 8900 }, { "epoch": 1.33, "learning_rate": 1.6333651535600252e-05, "loss": 0.4107, "step": 8950 }, { "epoch": 1.33, "learning_rate": 1.6300734059712304e-05, "loss": 0.4033, "step": 9000 }, { "epoch": 1.34, "learning_rate": 1.6267816583824355e-05, "loss": 0.3992, "step": 9050 }, { "epoch": 1.35, "learning_rate": 1.6234899107936402e-05, "loss": 0.3721, "step": 9100 }, { "epoch": 1.36, "learning_rate": 1.6201981632048457e-05, "loss": 0.3948, "step": 9150 }, { "epoch": 1.36, "learning_rate": 1.6169064156160508e-05, "loss": 0.4064, "step": 9200 }, { "epoch": 1.37, "learning_rate": 1.6136805029790316e-05, "loss": 0.4048, "step": 9250 }, { "epoch": 1.38, "learning_rate": 1.6103887553902367e-05, "loss": 0.3823, "step": 9300 }, { "epoch": 1.38, "learning_rate": 1.6070970078014418e-05, "loss": 0.3902, "step": 9350 }, { "epoch": 1.39, "learning_rate": 1.603805260212647e-05, "loss": 0.3738, "step": 9400 }, { "epoch": 1.4, "learning_rate": 1.6005135126238524e-05, "loss": 0.3811, "step": 9450 }, { "epoch": 1.41, "learning_rate": 1.597221765035057e-05, "loss": 0.3806, "step": 9500 }, { "epoch": 1.41, "learning_rate": 1.5939300174462626e-05, "loss": 0.3779, "step": 9550 }, { "epoch": 1.42, "learning_rate": 1.5906382698574673e-05, "loss": 0.38, "step": 9600 }, { "epoch": 1.43, "learning_rate": 1.5873465222686724e-05, "loss": 0.3967, "step": 9650 }, { "epoch": 1.44, "learning_rate": 1.584054774679878e-05, "loss": 0.3738, "step": 9700 }, { "epoch": 1.44, "learning_rate": 1.5807630270910826e-05, "loss": 0.3717, "step": 9750 }, { "epoch": 1.45, "learning_rate": 1.577471279502288e-05, "loss": 0.4036, "step": 9800 }, { "epoch": 1.46, "learning_rate": 1.574179531913493e-05, "loss": 0.3587, "step": 9850 }, { "epoch": 1.47, "learning_rate": 1.5708877843246983e-05, "loss": 0.3935, "step": 9900 }, { "epoch": 1.47, "learning_rate": 1.5675960367359034e-05, "loss": 0.3768, "step": 9950 }, { "epoch": 1.48, "learning_rate": 1.564304289147108e-05, "loss": 0.3848, "step": 10000 }, { "epoch": 1.48, "eval_loss": 0.16684819757938385, "eval_runtime": 93.9951, "eval_samples_per_second": 59.141, "eval_steps_per_second": 7.394, "eval_wer": 0.11944452387079557, "step": 10000 }, { "epoch": 1.49, "learning_rate": 1.5610125415583136e-05, "loss": 0.3784, "step": 10050 }, { "epoch": 1.5, "learning_rate": 1.5577207939695183e-05, "loss": 0.393, "step": 10100 }, { "epoch": 1.5, "learning_rate": 1.5544290463807238e-05, "loss": 0.3827, "step": 10150 }, { "epoch": 1.51, "learning_rate": 1.551137298791929e-05, "loss": 0.3781, "step": 10200 }, { "epoch": 1.52, "learning_rate": 1.547845551203134e-05, "loss": 0.3756, "step": 10250 }, { "epoch": 1.53, "learning_rate": 1.544553803614339e-05, "loss": 0.4135, "step": 10300 }, { "epoch": 1.53, "learning_rate": 1.541262056025544e-05, "loss": 0.38, "step": 10350 }, { "epoch": 1.54, "learning_rate": 1.5379703084367493e-05, "loss": 0.3689, "step": 10400 }, { "epoch": 1.55, "learning_rate": 1.5346785608479544e-05, "loss": 0.358, "step": 10450 }, { "epoch": 1.56, "learning_rate": 1.5313868132591595e-05, "loss": 0.387, "step": 10500 }, { "epoch": 1.56, "learning_rate": 1.5280950656703646e-05, "loss": 0.3673, "step": 10550 }, { "epoch": 1.57, "learning_rate": 1.5248033180815695e-05, "loss": 0.3579, "step": 10600 }, { "epoch": 1.58, "learning_rate": 1.5215115704927748e-05, "loss": 0.3909, "step": 10650 }, { "epoch": 1.58, "learning_rate": 1.5182198229039799e-05, "loss": 0.3702, "step": 10700 }, { "epoch": 1.59, "learning_rate": 1.514928075315185e-05, "loss": 0.395, "step": 10750 }, { "epoch": 1.6, "learning_rate": 1.5116363277263901e-05, "loss": 0.4106, "step": 10800 }, { "epoch": 1.61, "learning_rate": 1.508344580137595e-05, "loss": 0.4134, "step": 10850 }, { "epoch": 1.61, "learning_rate": 1.5050528325488003e-05, "loss": 0.3915, "step": 10900 }, { "epoch": 1.62, "learning_rate": 1.5017610849600056e-05, "loss": 0.3684, "step": 10950 }, { "epoch": 1.63, "learning_rate": 1.4984693373712105e-05, "loss": 0.3591, "step": 11000 }, { "epoch": 1.64, "learning_rate": 1.4951775897824156e-05, "loss": 0.3738, "step": 11050 }, { "epoch": 1.64, "learning_rate": 1.4918858421936207e-05, "loss": 0.3782, "step": 11100 }, { "epoch": 1.65, "learning_rate": 1.4885940946048258e-05, "loss": 0.3488, "step": 11150 }, { "epoch": 1.66, "learning_rate": 1.4853023470160311e-05, "loss": 0.3805, "step": 11200 }, { "epoch": 1.67, "learning_rate": 1.482010599427236e-05, "loss": 0.3627, "step": 11250 }, { "epoch": 1.67, "learning_rate": 1.4787188518384413e-05, "loss": 0.3828, "step": 11300 }, { "epoch": 1.68, "learning_rate": 1.4754271042496462e-05, "loss": 0.3473, "step": 11350 }, { "epoch": 1.69, "learning_rate": 1.4721353566608513e-05, "loss": 0.3789, "step": 11400 }, { "epoch": 1.7, "learning_rate": 1.4688436090720564e-05, "loss": 0.3718, "step": 11450 }, { "epoch": 1.7, "learning_rate": 1.4655518614832615e-05, "loss": 0.3576, "step": 11500 }, { "epoch": 1.71, "learning_rate": 1.4622601138944668e-05, "loss": 0.3623, "step": 11550 }, { "epoch": 1.72, "learning_rate": 1.4589683663056717e-05, "loss": 0.3604, "step": 11600 }, { "epoch": 1.73, "learning_rate": 1.455676618716877e-05, "loss": 0.3638, "step": 11650 }, { "epoch": 1.73, "learning_rate": 1.452384871128082e-05, "loss": 0.3571, "step": 11700 }, { "epoch": 1.74, "learning_rate": 1.449093123539287e-05, "loss": 0.3574, "step": 11750 }, { "epoch": 1.75, "learning_rate": 1.4458013759504923e-05, "loss": 0.3731, "step": 11800 }, { "epoch": 1.76, "learning_rate": 1.4425096283616972e-05, "loss": 0.3968, "step": 11850 }, { "epoch": 1.76, "learning_rate": 1.4392178807729025e-05, "loss": 0.3792, "step": 11900 }, { "epoch": 1.77, "learning_rate": 1.4359261331841074e-05, "loss": 0.3541, "step": 11950 }, { "epoch": 1.78, "learning_rate": 1.4326343855953127e-05, "loss": 0.3685, "step": 12000 }, { "epoch": 1.78, "learning_rate": 1.4293426380065178e-05, "loss": 0.3618, "step": 12050 }, { "epoch": 1.79, "learning_rate": 1.4260508904177227e-05, "loss": 0.3379, "step": 12100 }, { "epoch": 1.8, "learning_rate": 1.422759142828928e-05, "loss": 0.345, "step": 12150 }, { "epoch": 1.81, "learning_rate": 1.419467395240133e-05, "loss": 0.4049, "step": 12200 }, { "epoch": 1.81, "learning_rate": 1.4161756476513382e-05, "loss": 0.354, "step": 12250 }, { "epoch": 1.82, "learning_rate": 1.4128839000625433e-05, "loss": 0.3421, "step": 12300 }, { "epoch": 1.83, "learning_rate": 1.4095921524737484e-05, "loss": 0.3737, "step": 12350 }, { "epoch": 1.84, "learning_rate": 1.4063004048849535e-05, "loss": 0.3661, "step": 12400 }, { "epoch": 1.84, "learning_rate": 1.4030086572961585e-05, "loss": 0.3581, "step": 12450 }, { "epoch": 1.85, "learning_rate": 1.3997169097073637e-05, "loss": 0.3535, "step": 12500 }, { "epoch": 1.85, "eval_loss": 0.15184186398983002, "eval_runtime": 94.1778, "eval_samples_per_second": 59.027, "eval_steps_per_second": 7.38, "eval_wer": 0.10855040555094883, "step": 12500 }, { "epoch": 1.86, "learning_rate": 1.396425162118569e-05, "loss": 0.3545, "step": 12550 }, { "epoch": 1.87, "learning_rate": 1.393133414529774e-05, "loss": 0.3412, "step": 12600 }, { "epoch": 1.87, "learning_rate": 1.389841666940979e-05, "loss": 0.3584, "step": 12650 }, { "epoch": 1.88, "learning_rate": 1.3865499193521841e-05, "loss": 0.3448, "step": 12700 }, { "epoch": 1.89, "learning_rate": 1.3832581717633892e-05, "loss": 0.3345, "step": 12750 }, { "epoch": 1.9, "learning_rate": 1.3799664241745945e-05, "loss": 0.3611, "step": 12800 }, { "epoch": 1.9, "learning_rate": 1.3766746765857994e-05, "loss": 0.3221, "step": 12850 }, { "epoch": 1.91, "learning_rate": 1.3733829289970047e-05, "loss": 0.3586, "step": 12900 }, { "epoch": 1.92, "learning_rate": 1.3700911814082096e-05, "loss": 0.3341, "step": 12950 }, { "epoch": 1.93, "learning_rate": 1.3667994338194147e-05, "loss": 0.3476, "step": 13000 }, { "epoch": 1.93, "learning_rate": 1.36350768623062e-05, "loss": 0.356, "step": 13050 }, { "epoch": 1.94, "learning_rate": 1.360215938641825e-05, "loss": 0.3378, "step": 13100 }, { "epoch": 1.95, "learning_rate": 1.3569241910530302e-05, "loss": 0.3357, "step": 13150 }, { "epoch": 1.96, "learning_rate": 1.3536324434642351e-05, "loss": 0.3447, "step": 13200 }, { "epoch": 1.96, "learning_rate": 1.3503406958754404e-05, "loss": 0.3143, "step": 13250 }, { "epoch": 1.97, "learning_rate": 1.3470489482866455e-05, "loss": 0.3379, "step": 13300 }, { "epoch": 1.98, "learning_rate": 1.3437572006978506e-05, "loss": 0.3086, "step": 13350 }, { "epoch": 1.98, "learning_rate": 1.3404654531090557e-05, "loss": 0.361, "step": 13400 }, { "epoch": 1.99, "learning_rate": 1.3371737055202607e-05, "loss": 0.3202, "step": 13450 }, { "epoch": 2.0, "learning_rate": 1.333881957931466e-05, "loss": 0.3222, "step": 13500 }, { "epoch": 2.01, "learning_rate": 1.330590210342671e-05, "loss": 0.3063, "step": 13550 }, { "epoch": 2.01, "learning_rate": 1.3272984627538761e-05, "loss": 0.3444, "step": 13600 }, { "epoch": 2.02, "learning_rate": 1.3240067151650812e-05, "loss": 0.3146, "step": 13650 }, { "epoch": 2.03, "learning_rate": 1.3207149675762863e-05, "loss": 0.356, "step": 13700 }, { "epoch": 2.04, "learning_rate": 1.3174232199874914e-05, "loss": 0.338, "step": 13750 }, { "epoch": 2.04, "learning_rate": 1.3141314723986967e-05, "loss": 0.3471, "step": 13800 }, { "epoch": 2.05, "learning_rate": 1.3108397248099016e-05, "loss": 0.341, "step": 13850 }, { "epoch": 2.06, "learning_rate": 1.3075479772211069e-05, "loss": 0.3257, "step": 13900 }, { "epoch": 2.07, "learning_rate": 1.3042562296323118e-05, "loss": 0.3492, "step": 13950 }, { "epoch": 2.07, "learning_rate": 1.300964482043517e-05, "loss": 0.3373, "step": 14000 }, { "epoch": 2.08, "learning_rate": 1.2976727344547222e-05, "loss": 0.3587, "step": 14050 }, { "epoch": 2.09, "learning_rate": 1.2943809868659271e-05, "loss": 0.3527, "step": 14100 }, { "epoch": 2.1, "learning_rate": 1.2910892392771324e-05, "loss": 0.3293, "step": 14150 }, { "epoch": 2.1, "learning_rate": 1.2877974916883373e-05, "loss": 0.3471, "step": 14200 }, { "epoch": 2.11, "learning_rate": 1.2845057440995426e-05, "loss": 0.3302, "step": 14250 }, { "epoch": 2.12, "learning_rate": 1.2812139965107477e-05, "loss": 0.3359, "step": 14300 }, { "epoch": 2.13, "learning_rate": 1.2779222489219527e-05, "loss": 0.3387, "step": 14350 }, { "epoch": 2.13, "learning_rate": 1.274630501333158e-05, "loss": 0.3532, "step": 14400 }, { "epoch": 2.14, "learning_rate": 1.2713387537443629e-05, "loss": 0.3234, "step": 14450 }, { "epoch": 2.15, "learning_rate": 1.2680470061555681e-05, "loss": 0.3082, "step": 14500 }, { "epoch": 2.16, "learning_rate": 1.2647552585667732e-05, "loss": 0.341, "step": 14550 }, { "epoch": 2.16, "learning_rate": 1.2614635109779783e-05, "loss": 0.3458, "step": 14600 }, { "epoch": 2.17, "learning_rate": 1.2581717633891834e-05, "loss": 0.3533, "step": 14650 }, { "epoch": 2.18, "learning_rate": 1.2548800158003884e-05, "loss": 0.3091, "step": 14700 }, { "epoch": 2.18, "learning_rate": 1.2515882682115936e-05, "loss": 0.3265, "step": 14750 }, { "epoch": 2.19, "learning_rate": 1.2482965206227989e-05, "loss": 0.3204, "step": 14800 }, { "epoch": 2.2, "learning_rate": 1.2450047730340038e-05, "loss": 0.326, "step": 14850 }, { "epoch": 2.21, "learning_rate": 1.241713025445209e-05, "loss": 0.3381, "step": 14900 }, { "epoch": 2.21, "learning_rate": 1.238421277856414e-05, "loss": 0.3321, "step": 14950 }, { "epoch": 2.22, "learning_rate": 1.2351295302676191e-05, "loss": 0.3667, "step": 15000 }, { "epoch": 2.22, "eval_loss": 0.1441650092601776, "eval_runtime": 94.5951, "eval_samples_per_second": 58.766, "eval_steps_per_second": 7.347, "eval_wer": 0.10192624786740247, "step": 15000 }, { "epoch": 2.23, "learning_rate": 1.2318377826788244e-05, "loss": 0.295, "step": 15050 }, { "epoch": 2.24, "learning_rate": 1.2285460350900293e-05, "loss": 0.3484, "step": 15100 }, { "epoch": 2.24, "learning_rate": 1.2252542875012346e-05, "loss": 0.2943, "step": 15150 }, { "epoch": 2.25, "learning_rate": 1.2219625399124395e-05, "loss": 0.3193, "step": 15200 }, { "epoch": 2.26, "learning_rate": 1.2187366272754206e-05, "loss": 0.3237, "step": 15250 }, { "epoch": 2.27, "learning_rate": 1.2154448796866259e-05, "loss": 0.3266, "step": 15300 }, { "epoch": 2.27, "learning_rate": 1.2121531320978308e-05, "loss": 0.3084, "step": 15350 }, { "epoch": 2.28, "learning_rate": 1.208861384509036e-05, "loss": 0.3041, "step": 15400 }, { "epoch": 2.29, "learning_rate": 1.205569636920241e-05, "loss": 0.314, "step": 15450 }, { "epoch": 2.3, "learning_rate": 1.2022778893314463e-05, "loss": 0.3343, "step": 15500 }, { "epoch": 2.3, "learning_rate": 1.1989861417426514e-05, "loss": 0.3128, "step": 15550 }, { "epoch": 2.31, "learning_rate": 1.1956943941538563e-05, "loss": 0.3197, "step": 15600 }, { "epoch": 2.32, "learning_rate": 1.1924026465650616e-05, "loss": 0.3098, "step": 15650 }, { "epoch": 2.33, "learning_rate": 1.1891108989762665e-05, "loss": 0.3152, "step": 15700 }, { "epoch": 2.33, "learning_rate": 1.1858191513874718e-05, "loss": 0.3346, "step": 15750 }, { "epoch": 2.34, "learning_rate": 1.1825274037986769e-05, "loss": 0.3083, "step": 15800 }, { "epoch": 2.35, "learning_rate": 1.179235656209882e-05, "loss": 0.3158, "step": 15850 }, { "epoch": 2.36, "learning_rate": 1.175943908621087e-05, "loss": 0.3269, "step": 15900 }, { "epoch": 2.36, "learning_rate": 1.172652161032292e-05, "loss": 0.322, "step": 15950 }, { "epoch": 2.37, "learning_rate": 1.1693604134434973e-05, "loss": 0.2992, "step": 16000 }, { "epoch": 2.38, "learning_rate": 1.1660686658547025e-05, "loss": 0.2987, "step": 16050 }, { "epoch": 2.38, "learning_rate": 1.1627769182659075e-05, "loss": 0.3072, "step": 16100 }, { "epoch": 2.39, "learning_rate": 1.1594851706771126e-05, "loss": 0.3153, "step": 16150 }, { "epoch": 2.4, "learning_rate": 1.1561934230883177e-05, "loss": 0.3106, "step": 16200 }, { "epoch": 2.41, "learning_rate": 1.1529016754995228e-05, "loss": 0.2987, "step": 16250 }, { "epoch": 2.41, "learning_rate": 1.149609927910728e-05, "loss": 0.3052, "step": 16300 }, { "epoch": 2.42, "learning_rate": 1.146318180321933e-05, "loss": 0.3192, "step": 16350 }, { "epoch": 2.43, "learning_rate": 1.1430264327331383e-05, "loss": 0.3612, "step": 16400 }, { "epoch": 2.44, "learning_rate": 1.1397346851443432e-05, "loss": 0.3007, "step": 16450 }, { "epoch": 2.44, "learning_rate": 1.1364429375555483e-05, "loss": 0.3289, "step": 16500 }, { "epoch": 2.45, "learning_rate": 1.1331511899667536e-05, "loss": 0.3201, "step": 16550 }, { "epoch": 2.46, "learning_rate": 1.1298594423779585e-05, "loss": 0.3283, "step": 16600 }, { "epoch": 2.47, "learning_rate": 1.1265676947891638e-05, "loss": 0.2926, "step": 16650 }, { "epoch": 2.47, "learning_rate": 1.1232759472003687e-05, "loss": 0.3256, "step": 16700 }, { "epoch": 2.48, "learning_rate": 1.119984199611574e-05, "loss": 0.3033, "step": 16750 }, { "epoch": 2.49, "learning_rate": 1.116692452022779e-05, "loss": 0.2949, "step": 16800 }, { "epoch": 2.5, "learning_rate": 1.113400704433984e-05, "loss": 0.3104, "step": 16850 }, { "epoch": 2.5, "learning_rate": 1.1101089568451893e-05, "loss": 0.3191, "step": 16900 }, { "epoch": 2.51, "learning_rate": 1.1068172092563942e-05, "loss": 0.3158, "step": 16950 }, { "epoch": 2.52, "learning_rate": 1.1035254616675995e-05, "loss": 0.2887, "step": 17000 }, { "epoch": 2.53, "learning_rate": 1.1002337140788046e-05, "loss": 0.3118, "step": 17050 }, { "epoch": 2.53, "learning_rate": 1.0969419664900097e-05, "loss": 0.3035, "step": 17100 }, { "epoch": 2.54, "learning_rate": 1.0936502189012148e-05, "loss": 0.2799, "step": 17150 }, { "epoch": 2.55, "learning_rate": 1.0903584713124197e-05, "loss": 0.2985, "step": 17200 }, { "epoch": 2.56, "learning_rate": 1.0871325586754009e-05, "loss": 0.316, "step": 17250 }, { "epoch": 2.56, "learning_rate": 1.0838408110866058e-05, "loss": 0.3301, "step": 17300 }, { "epoch": 2.57, "learning_rate": 1.0805490634978111e-05, "loss": 0.2964, "step": 17350 }, { "epoch": 2.58, "learning_rate": 1.0772573159090162e-05, "loss": 0.3055, "step": 17400 }, { "epoch": 2.58, "learning_rate": 1.0739655683202213e-05, "loss": 0.3069, "step": 17450 }, { "epoch": 2.59, "learning_rate": 1.0706738207314264e-05, "loss": 0.3058, "step": 17500 }, { "epoch": 2.59, "eval_loss": 0.13805590569972992, "eval_runtime": 94.4292, "eval_samples_per_second": 58.869, "eval_steps_per_second": 7.36, "eval_wer": 0.09607411431675865, "step": 17500 }, { "epoch": 2.6, "learning_rate": 1.0673820731426313e-05, "loss": 0.2935, "step": 17550 }, { "epoch": 2.61, "learning_rate": 1.0640903255538366e-05, "loss": 0.3076, "step": 17600 }, { "epoch": 2.61, "learning_rate": 1.0607985779650419e-05, "loss": 0.3513, "step": 17650 }, { "epoch": 2.62, "learning_rate": 1.0575068303762468e-05, "loss": 0.3112, "step": 17700 }, { "epoch": 2.63, "learning_rate": 1.054215082787452e-05, "loss": 0.3216, "step": 17750 }, { "epoch": 2.64, "learning_rate": 1.050923335198657e-05, "loss": 0.3259, "step": 17800 }, { "epoch": 2.64, "learning_rate": 1.0476315876098621e-05, "loss": 0.3283, "step": 17850 }, { "epoch": 2.65, "learning_rate": 1.0443398400210674e-05, "loss": 0.3189, "step": 17900 }, { "epoch": 2.66, "learning_rate": 1.0410480924322723e-05, "loss": 0.3099, "step": 17950 }, { "epoch": 2.67, "learning_rate": 1.0377563448434776e-05, "loss": 0.3049, "step": 18000 }, { "epoch": 2.67, "learning_rate": 1.0344645972546825e-05, "loss": 0.3227, "step": 18050 }, { "epoch": 2.68, "learning_rate": 1.0311728496658876e-05, "loss": 0.2857, "step": 18100 }, { "epoch": 2.69, "learning_rate": 1.0278811020770929e-05, "loss": 0.3148, "step": 18150 }, { "epoch": 2.7, "learning_rate": 1.0245893544882978e-05, "loss": 0.3378, "step": 18200 }, { "epoch": 2.7, "learning_rate": 1.0212976068995031e-05, "loss": 0.3206, "step": 18250 }, { "epoch": 2.71, "learning_rate": 1.018005859310708e-05, "loss": 0.308, "step": 18300 }, { "epoch": 2.72, "learning_rate": 1.0147141117219133e-05, "loss": 0.298, "step": 18350 }, { "epoch": 2.73, "learning_rate": 1.0114223641331184e-05, "loss": 0.3179, "step": 18400 }, { "epoch": 2.73, "learning_rate": 1.0081306165443233e-05, "loss": 0.2943, "step": 18450 }, { "epoch": 2.74, "learning_rate": 1.0048388689555286e-05, "loss": 0.3023, "step": 18500 }, { "epoch": 2.75, "learning_rate": 1.0015471213667335e-05, "loss": 0.2826, "step": 18550 }, { "epoch": 2.76, "learning_rate": 9.982553737779388e-06, "loss": 0.3004, "step": 18600 }, { "epoch": 2.76, "learning_rate": 9.94963626189144e-06, "loss": 0.325, "step": 18650 }, { "epoch": 2.77, "learning_rate": 9.91671878600349e-06, "loss": 0.302, "step": 18700 }, { "epoch": 2.78, "learning_rate": 9.883801310115541e-06, "loss": 0.2962, "step": 18750 }, { "epoch": 2.78, "learning_rate": 9.850883834227592e-06, "loss": 0.2967, "step": 18800 }, { "epoch": 2.79, "learning_rate": 9.817966358339643e-06, "loss": 0.2756, "step": 18850 }, { "epoch": 2.8, "learning_rate": 9.785048882451694e-06, "loss": 0.2829, "step": 18900 }, { "epoch": 2.81, "learning_rate": 9.752131406563745e-06, "loss": 0.3034, "step": 18950 }, { "epoch": 2.81, "learning_rate": 9.719213930675796e-06, "loss": 0.3011, "step": 19000 }, { "epoch": 2.82, "learning_rate": 9.686296454787847e-06, "loss": 0.296, "step": 19050 }, { "epoch": 2.83, "learning_rate": 9.653378978899898e-06, "loss": 0.3338, "step": 19100 }, { "epoch": 2.84, "learning_rate": 9.62046150301195e-06, "loss": 0.2975, "step": 19150 }, { "epoch": 2.84, "learning_rate": 9.587544027124e-06, "loss": 0.3063, "step": 19200 }, { "epoch": 2.85, "learning_rate": 9.554626551236053e-06, "loss": 0.3165, "step": 19250 }, { "epoch": 2.86, "learning_rate": 9.521709075348104e-06, "loss": 0.3146, "step": 19300 }, { "epoch": 2.87, "learning_rate": 9.488791599460155e-06, "loss": 0.274, "step": 19350 }, { "epoch": 2.87, "learning_rate": 9.455874123572204e-06, "loss": 0.3061, "step": 19400 }, { "epoch": 2.88, "learning_rate": 9.422956647684255e-06, "loss": 0.2878, "step": 19450 }, { "epoch": 2.89, "learning_rate": 9.390697521314066e-06, "loss": 0.3059, "step": 19500 }, { "epoch": 2.9, "learning_rate": 9.357780045426117e-06, "loss": 0.2902, "step": 19550 }, { "epoch": 2.9, "learning_rate": 9.32486256953817e-06, "loss": 0.316, "step": 19600 }, { "epoch": 2.91, "learning_rate": 9.29194509365022e-06, "loss": 0.2864, "step": 19650 }, { "epoch": 2.92, "learning_rate": 9.25902761776227e-06, "loss": 0.3141, "step": 19700 }, { "epoch": 2.93, "learning_rate": 9.226110141874321e-06, "loss": 0.2801, "step": 19750 }, { "epoch": 2.93, "learning_rate": 9.193192665986372e-06, "loss": 0.2827, "step": 19800 }, { "epoch": 2.94, "learning_rate": 9.160275190098425e-06, "loss": 0.3031, "step": 19850 }, { "epoch": 2.95, "learning_rate": 9.127357714210476e-06, "loss": 0.2879, "step": 19900 }, { "epoch": 2.96, "learning_rate": 9.094440238322527e-06, "loss": 0.303, "step": 19950 }, { "epoch": 2.96, "learning_rate": 9.061522762434578e-06, "loss": 0.3026, "step": 20000 }, { "epoch": 2.96, "eval_loss": 0.1326620876789093, "eval_runtime": 94.0377, "eval_samples_per_second": 59.115, "eval_steps_per_second": 7.391, "eval_wer": 0.092404616894938, "step": 20000 }, { "epoch": 2.97, "learning_rate": 9.028605286546629e-06, "loss": 0.3161, "step": 20050 }, { "epoch": 2.98, "learning_rate": 8.99568781065868e-06, "loss": 0.2733, "step": 20100 }, { "epoch": 2.98, "learning_rate": 8.96277033477073e-06, "loss": 0.3036, "step": 20150 }, { "epoch": 2.99, "learning_rate": 8.929852858882782e-06, "loss": 0.277, "step": 20200 }, { "epoch": 3.0, "learning_rate": 8.896935382994833e-06, "loss": 0.2755, "step": 20250 }, { "epoch": 3.01, "learning_rate": 8.864017907106884e-06, "loss": 0.2808, "step": 20300 }, { "epoch": 3.01, "learning_rate": 8.831100431218935e-06, "loss": 0.3029, "step": 20350 }, { "epoch": 3.02, "learning_rate": 8.798182955330986e-06, "loss": 0.315, "step": 20400 }, { "epoch": 3.03, "learning_rate": 8.765265479443037e-06, "loss": 0.3076, "step": 20450 }, { "epoch": 3.04, "learning_rate": 8.732348003555088e-06, "loss": 0.3095, "step": 20500 }, { "epoch": 3.04, "learning_rate": 8.699430527667139e-06, "loss": 0.3107, "step": 20550 }, { "epoch": 3.05, "learning_rate": 8.666513051779192e-06, "loss": 0.2882, "step": 20600 }, { "epoch": 3.06, "learning_rate": 8.63359557589124e-06, "loss": 0.2871, "step": 20650 }, { "epoch": 3.07, "learning_rate": 8.600678100003292e-06, "loss": 0.3163, "step": 20700 }, { "epoch": 3.07, "learning_rate": 8.567760624115343e-06, "loss": 0.3136, "step": 20750 }, { "epoch": 3.08, "learning_rate": 8.534843148227394e-06, "loss": 0.2999, "step": 20800 }, { "epoch": 3.09, "learning_rate": 8.501925672339447e-06, "loss": 0.2836, "step": 20850 }, { "epoch": 3.1, "learning_rate": 8.469008196451498e-06, "loss": 0.3043, "step": 20900 }, { "epoch": 3.1, "learning_rate": 8.436090720563549e-06, "loss": 0.2978, "step": 20950 }, { "epoch": 3.11, "learning_rate": 8.403173244675598e-06, "loss": 0.3028, "step": 21000 }, { "epoch": 3.12, "learning_rate": 8.370255768787649e-06, "loss": 0.3079, "step": 21050 }, { "epoch": 3.13, "learning_rate": 8.337338292899702e-06, "loss": 0.2724, "step": 21100 }, { "epoch": 3.13, "learning_rate": 8.304420817011753e-06, "loss": 0.3063, "step": 21150 }, { "epoch": 3.14, "learning_rate": 8.271503341123804e-06, "loss": 0.2956, "step": 21200 }, { "epoch": 3.15, "learning_rate": 8.238585865235855e-06, "loss": 0.2925, "step": 21250 }, { "epoch": 3.16, "learning_rate": 8.205668389347906e-06, "loss": 0.2796, "step": 21300 }, { "epoch": 3.16, "learning_rate": 8.172750913459957e-06, "loss": 0.2867, "step": 21350 }, { "epoch": 3.17, "learning_rate": 8.139833437572008e-06, "loss": 0.2905, "step": 21400 }, { "epoch": 3.18, "learning_rate": 8.106915961684059e-06, "loss": 0.3393, "step": 21450 }, { "epoch": 3.18, "learning_rate": 8.07399848579611e-06, "loss": 0.3006, "step": 21500 }, { "epoch": 3.19, "learning_rate": 8.04108100990816e-06, "loss": 0.2797, "step": 21550 }, { "epoch": 3.2, "learning_rate": 8.008163534020212e-06, "loss": 0.2569, "step": 21600 }, { "epoch": 3.21, "learning_rate": 7.975246058132263e-06, "loss": 0.2988, "step": 21650 }, { "epoch": 3.21, "learning_rate": 7.942328582244314e-06, "loss": 0.2782, "step": 21700 }, { "epoch": 3.22, "learning_rate": 7.909411106356365e-06, "loss": 0.2972, "step": 21750 }, { "epoch": 3.23, "learning_rate": 7.876493630468416e-06, "loss": 0.3061, "step": 21800 }, { "epoch": 3.24, "learning_rate": 7.843576154580469e-06, "loss": 0.2716, "step": 21850 }, { "epoch": 3.24, "learning_rate": 7.810658678692518e-06, "loss": 0.2931, "step": 21900 }, { "epoch": 3.25, "learning_rate": 7.777741202804569e-06, "loss": 0.3147, "step": 21950 }, { "epoch": 3.26, "learning_rate": 7.74482372691662e-06, "loss": 0.3011, "step": 22000 }, { "epoch": 3.27, "learning_rate": 7.711906251028671e-06, "loss": 0.3163, "step": 22050 }, { "epoch": 3.27, "learning_rate": 7.678988775140724e-06, "loss": 0.2816, "step": 22100 }, { "epoch": 3.28, "learning_rate": 7.646071299252775e-06, "loss": 0.2824, "step": 22150 }, { "epoch": 3.29, "learning_rate": 7.613153823364825e-06, "loss": 0.3133, "step": 22200 }, { "epoch": 3.3, "learning_rate": 7.580236347476876e-06, "loss": 0.2905, "step": 22250 }, { "epoch": 3.3, "learning_rate": 7.547318871588927e-06, "loss": 0.2808, "step": 22300 }, { "epoch": 3.31, "learning_rate": 7.514401395700979e-06, "loss": 0.3104, "step": 22350 }, { "epoch": 3.32, "learning_rate": 7.48148391981303e-06, "loss": 0.2931, "step": 22400 }, { "epoch": 3.33, "learning_rate": 7.448566443925081e-06, "loss": 0.2998, "step": 22450 }, { "epoch": 3.33, "learning_rate": 7.415648968037131e-06, "loss": 0.2891, "step": 22500 }, { "epoch": 3.33, "eval_loss": 0.13261927664279938, "eval_runtime": 94.2326, "eval_samples_per_second": 58.992, "eval_steps_per_second": 7.375, "eval_wer": 0.09173743554551607, "step": 22500 }, { "epoch": 3.34, "learning_rate": 7.382731492149182e-06, "loss": 0.305, "step": 22550 }, { "epoch": 3.35, "learning_rate": 7.349814016261234e-06, "loss": 0.3068, "step": 22600 }, { "epoch": 3.36, "learning_rate": 7.316896540373285e-06, "loss": 0.2669, "step": 22650 }, { "epoch": 3.36, "learning_rate": 7.283979064485336e-06, "loss": 0.2682, "step": 22700 }, { "epoch": 3.37, "learning_rate": 7.251061588597387e-06, "loss": 0.3007, "step": 22750 }, { "epoch": 3.38, "learning_rate": 7.218144112709438e-06, "loss": 0.267, "step": 22800 }, { "epoch": 3.38, "learning_rate": 7.18522663682149e-06, "loss": 0.2941, "step": 22850 }, { "epoch": 3.39, "learning_rate": 7.1529675104513e-06, "loss": 0.2855, "step": 22900 }, { "epoch": 3.4, "learning_rate": 7.12005003456335e-06, "loss": 0.2661, "step": 22950 }, { "epoch": 3.41, "learning_rate": 7.087132558675401e-06, "loss": 0.2959, "step": 23000 }, { "epoch": 3.41, "learning_rate": 7.054215082787452e-06, "loss": 0.2864, "step": 23050 }, { "epoch": 3.42, "learning_rate": 7.021297606899503e-06, "loss": 0.301, "step": 23100 }, { "epoch": 3.43, "learning_rate": 6.988380131011555e-06, "loss": 0.2909, "step": 23150 }, { "epoch": 3.44, "learning_rate": 6.955462655123606e-06, "loss": 0.3021, "step": 23200 }, { "epoch": 3.44, "learning_rate": 6.922545179235657e-06, "loss": 0.2611, "step": 23250 }, { "epoch": 3.45, "learning_rate": 6.889627703347707e-06, "loss": 0.2972, "step": 23300 }, { "epoch": 3.46, "learning_rate": 6.856710227459758e-06, "loss": 0.2787, "step": 23350 }, { "epoch": 3.47, "learning_rate": 6.82379275157181e-06, "loss": 0.2687, "step": 23400 }, { "epoch": 3.47, "learning_rate": 6.790875275683861e-06, "loss": 0.2829, "step": 23450 }, { "epoch": 3.48, "learning_rate": 6.757957799795912e-06, "loss": 0.2785, "step": 23500 }, { "epoch": 3.49, "learning_rate": 6.725040323907963e-06, "loss": 0.3306, "step": 23550 }, { "epoch": 3.5, "learning_rate": 6.692122848020014e-06, "loss": 0.2982, "step": 23600 }, { "epoch": 3.5, "learning_rate": 6.659205372132066e-06, "loss": 0.2841, "step": 23650 }, { "epoch": 3.51, "learning_rate": 6.626287896244117e-06, "loss": 0.2541, "step": 23700 }, { "epoch": 3.52, "learning_rate": 6.593370420356167e-06, "loss": 0.2619, "step": 23750 }, { "epoch": 3.53, "learning_rate": 6.560452944468218e-06, "loss": 0.2777, "step": 23800 }, { "epoch": 3.53, "learning_rate": 6.527535468580269e-06, "loss": 0.2595, "step": 23850 }, { "epoch": 3.54, "learning_rate": 6.494617992692321e-06, "loss": 0.281, "step": 23900 }, { "epoch": 3.55, "learning_rate": 6.461700516804372e-06, "loss": 0.2974, "step": 23950 }, { "epoch": 3.56, "learning_rate": 6.428783040916423e-06, "loss": 0.2942, "step": 24000 }, { "epoch": 3.56, "learning_rate": 6.395865565028474e-06, "loss": 0.3043, "step": 24050 }, { "epoch": 3.57, "learning_rate": 6.362948089140525e-06, "loss": 0.2866, "step": 24100 }, { "epoch": 3.58, "learning_rate": 6.330030613252577e-06, "loss": 0.2838, "step": 24150 }, { "epoch": 3.58, "learning_rate": 6.297113137364627e-06, "loss": 0.2918, "step": 24200 }, { "epoch": 3.59, "learning_rate": 6.264195661476678e-06, "loss": 0.2721, "step": 24250 }, { "epoch": 3.6, "learning_rate": 6.231278185588729e-06, "loss": 0.2722, "step": 24300 }, { "epoch": 3.61, "learning_rate": 6.19836070970078e-06, "loss": 0.2807, "step": 24350 }, { "epoch": 3.61, "learning_rate": 6.165443233812832e-06, "loss": 0.31, "step": 24400 }, { "epoch": 3.62, "learning_rate": 6.132525757924883e-06, "loss": 0.3111, "step": 24450 }, { "epoch": 3.63, "learning_rate": 6.099608282036934e-06, "loss": 0.2896, "step": 24500 }, { "epoch": 3.64, "learning_rate": 6.066690806148985e-06, "loss": 0.288, "step": 24550 }, { "epoch": 3.64, "learning_rate": 6.033773330261035e-06, "loss": 0.2712, "step": 24600 }, { "epoch": 3.65, "learning_rate": 6.000855854373088e-06, "loss": 0.2938, "step": 24650 }, { "epoch": 3.66, "learning_rate": 5.967938378485138e-06, "loss": 0.2993, "step": 24700 }, { "epoch": 3.67, "learning_rate": 5.935020902597189e-06, "loss": 0.2818, "step": 24750 }, { "epoch": 3.67, "learning_rate": 5.90210342670924e-06, "loss": 0.2627, "step": 24800 }, { "epoch": 3.68, "learning_rate": 5.869185950821291e-06, "loss": 0.27, "step": 24850 }, { "epoch": 3.69, "learning_rate": 5.836268474933343e-06, "loss": 0.2664, "step": 24900 }, { "epoch": 3.7, "learning_rate": 5.803350999045394e-06, "loss": 0.2904, "step": 24950 }, { "epoch": 3.7, "learning_rate": 5.770433523157445e-06, "loss": 0.294, "step": 25000 }, { "epoch": 3.7, "eval_loss": 0.12781964242458344, "eval_runtime": 94.2202, "eval_samples_per_second": 59.0, "eval_steps_per_second": 7.376, "eval_wer": 0.08942136314680849, "step": 25000 }, { "epoch": 3.71, "learning_rate": 5.737516047269495e-06, "loss": 0.2902, "step": 25050 }, { "epoch": 3.72, "learning_rate": 5.704598571381546e-06, "loss": 0.2639, "step": 25100 }, { "epoch": 3.73, "learning_rate": 5.671681095493597e-06, "loss": 0.2746, "step": 25150 }, { "epoch": 3.73, "learning_rate": 5.638763619605649e-06, "loss": 0.311, "step": 25200 }, { "epoch": 3.74, "learning_rate": 5.6058461437177e-06, "loss": 0.306, "step": 25250 }, { "epoch": 3.75, "learning_rate": 5.572928667829751e-06, "loss": 0.2731, "step": 25300 }, { "epoch": 3.75, "learning_rate": 5.540011191941802e-06, "loss": 0.2866, "step": 25350 }, { "epoch": 3.76, "learning_rate": 5.5070937160538525e-06, "loss": 0.2517, "step": 25400 }, { "epoch": 3.77, "learning_rate": 5.474176240165905e-06, "loss": 0.2776, "step": 25450 }, { "epoch": 3.78, "learning_rate": 5.441258764277955e-06, "loss": 0.2771, "step": 25500 }, { "epoch": 3.78, "learning_rate": 5.408341288390006e-06, "loss": 0.286, "step": 25550 }, { "epoch": 3.79, "learning_rate": 5.375423812502057e-06, "loss": 0.2897, "step": 25600 }, { "epoch": 3.8, "learning_rate": 5.342506336614108e-06, "loss": 0.2796, "step": 25650 }, { "epoch": 3.81, "learning_rate": 5.30958886072616e-06, "loss": 0.2954, "step": 25700 }, { "epoch": 3.81, "learning_rate": 5.276671384838211e-06, "loss": 0.2653, "step": 25750 }, { "epoch": 3.82, "learning_rate": 5.243753908950262e-06, "loss": 0.2871, "step": 25800 }, { "epoch": 3.83, "learning_rate": 5.210836433062313e-06, "loss": 0.2814, "step": 25850 }, { "epoch": 3.84, "learning_rate": 5.1779189571743635e-06, "loss": 0.308, "step": 25900 }, { "epoch": 3.84, "learning_rate": 5.145001481286415e-06, "loss": 0.2577, "step": 25950 }, { "epoch": 3.85, "learning_rate": 5.112084005398466e-06, "loss": 0.2755, "step": 26000 }, { "epoch": 3.86, "learning_rate": 5.079166529510517e-06, "loss": 0.307, "step": 26050 }, { "epoch": 3.87, "learning_rate": 5.046249053622568e-06, "loss": 0.2729, "step": 26100 }, { "epoch": 3.87, "learning_rate": 5.013331577734619e-06, "loss": 0.27, "step": 26150 }, { "epoch": 3.88, "learning_rate": 4.9804141018466704e-06, "loss": 0.2789, "step": 26200 }, { "epoch": 3.89, "learning_rate": 4.947496625958722e-06, "loss": 0.2934, "step": 26250 }, { "epoch": 3.9, "learning_rate": 4.914579150070773e-06, "loss": 0.2699, "step": 26300 }, { "epoch": 3.9, "learning_rate": 4.881661674182824e-06, "loss": 0.3135, "step": 26350 }, { "epoch": 3.91, "learning_rate": 4.848744198294875e-06, "loss": 0.2837, "step": 26400 }, { "epoch": 3.92, "learning_rate": 4.815826722406926e-06, "loss": 0.2879, "step": 26450 }, { "epoch": 3.93, "learning_rate": 4.782909246518977e-06, "loss": 0.2791, "step": 26500 }, { "epoch": 3.93, "learning_rate": 4.749991770631028e-06, "loss": 0.2871, "step": 26550 }, { "epoch": 3.94, "learning_rate": 4.717074294743079e-06, "loss": 0.2793, "step": 26600 }, { "epoch": 3.95, "learning_rate": 4.68415681885513e-06, "loss": 0.2943, "step": 26650 }, { "epoch": 3.95, "learning_rate": 4.651239342967181e-06, "loss": 0.2947, "step": 26700 }, { "epoch": 3.96, "learning_rate": 4.618321867079233e-06, "loss": 0.265, "step": 26750 }, { "epoch": 3.97, "learning_rate": 4.5854043911912835e-06, "loss": 0.2829, "step": 26800 }, { "epoch": 3.98, "learning_rate": 4.5524869153033345e-06, "loss": 0.2756, "step": 26850 }, { "epoch": 3.98, "learning_rate": 4.520227788933145e-06, "loss": 0.2775, "step": 26900 }, { "epoch": 3.99, "learning_rate": 4.487310313045196e-06, "loss": 0.2786, "step": 26950 }, { "epoch": 4.0, "learning_rate": 4.454392837157247e-06, "loss": 0.2803, "step": 27000 }, { "epoch": 4.01, "learning_rate": 4.421475361269299e-06, "loss": 0.2618, "step": 27050 }, { "epoch": 4.01, "learning_rate": 4.38855788538135e-06, "loss": 0.2509, "step": 27100 }, { "epoch": 4.02, "learning_rate": 4.356298759011159e-06, "loss": 0.2763, "step": 27150 }, { "epoch": 4.03, "learning_rate": 4.32338128312321e-06, "loss": 0.2744, "step": 27200 }, { "epoch": 4.04, "learning_rate": 4.290463807235262e-06, "loss": 0.2785, "step": 27250 }, { "epoch": 4.04, "learning_rate": 4.257546331347312e-06, "loss": 0.2874, "step": 27300 }, { "epoch": 4.05, "learning_rate": 4.224628855459364e-06, "loss": 0.2801, "step": 27350 }, { "epoch": 4.06, "learning_rate": 4.191711379571415e-06, "loss": 0.2678, "step": 27400 }, { "epoch": 4.07, "learning_rate": 4.158793903683465e-06, "loss": 0.2764, "step": 27450 }, { "epoch": 4.07, "learning_rate": 4.125876427795517e-06, "loss": 0.2846, "step": 27500 }, { "epoch": 4.07, "eval_loss": 0.12571895122528076, "eval_runtime": 94.4519, "eval_samples_per_second": 58.855, "eval_steps_per_second": 7.358, "eval_wer": 0.08847777809548318, "step": 27500 }, { "epoch": 4.08, "learning_rate": 4.092958951907568e-06, "loss": 0.2553, "step": 27550 }, { "epoch": 4.09, "learning_rate": 4.060041476019619e-06, "loss": 0.2612, "step": 27600 }, { "epoch": 4.1, "learning_rate": 4.02712400013167e-06, "loss": 0.2504, "step": 27650 }, { "epoch": 4.1, "learning_rate": 3.994206524243721e-06, "loss": 0.2717, "step": 27700 }, { "epoch": 4.11, "learning_rate": 3.961289048355772e-06, "loss": 0.2738, "step": 27750 }, { "epoch": 4.12, "learning_rate": 3.928371572467823e-06, "loss": 0.285, "step": 27800 }, { "epoch": 4.13, "learning_rate": 3.895454096579875e-06, "loss": 0.2947, "step": 27850 }, { "epoch": 4.13, "learning_rate": 3.862536620691926e-06, "loss": 0.2676, "step": 27900 }, { "epoch": 4.14, "learning_rate": 3.829619144803976e-06, "loss": 0.2773, "step": 27950 }, { "epoch": 4.15, "learning_rate": 3.796701668916028e-06, "loss": 0.2805, "step": 28000 }, { "epoch": 4.15, "learning_rate": 3.7637841930280787e-06, "loss": 0.2691, "step": 28050 }, { "epoch": 4.16, "learning_rate": 3.73086671714013e-06, "loss": 0.2955, "step": 28100 }, { "epoch": 4.17, "learning_rate": 3.697949241252181e-06, "loss": 0.2676, "step": 28150 }, { "epoch": 4.18, "learning_rate": 3.665031765364232e-06, "loss": 0.278, "step": 28200 }, { "epoch": 4.18, "learning_rate": 3.6321142894762832e-06, "loss": 0.2655, "step": 28250 }, { "epoch": 4.19, "learning_rate": 3.5991968135883342e-06, "loss": 0.274, "step": 28300 }, { "epoch": 4.2, "learning_rate": 3.5662793377003857e-06, "loss": 0.2809, "step": 28350 }, { "epoch": 4.21, "learning_rate": 3.5333618618124367e-06, "loss": 0.265, "step": 28400 }, { "epoch": 4.21, "learning_rate": 3.5004443859244873e-06, "loss": 0.3473, "step": 28450 }, { "epoch": 4.22, "learning_rate": 3.4675269100365387e-06, "loss": 0.2519, "step": 28500 }, { "epoch": 4.23, "learning_rate": 3.4346094341485897e-06, "loss": 0.283, "step": 28550 }, { "epoch": 4.24, "learning_rate": 3.4016919582606407e-06, "loss": 0.2998, "step": 28600 }, { "epoch": 4.24, "learning_rate": 3.368774482372692e-06, "loss": 0.2855, "step": 28650 }, { "epoch": 4.25, "learning_rate": 3.3358570064847428e-06, "loss": 0.2829, "step": 28700 }, { "epoch": 4.26, "learning_rate": 3.3029395305967942e-06, "loss": 0.2742, "step": 28750 }, { "epoch": 4.27, "learning_rate": 3.2700220547088452e-06, "loss": 0.2943, "step": 28800 }, { "epoch": 4.27, "learning_rate": 3.237104578820896e-06, "loss": 0.2748, "step": 28850 }, { "epoch": 4.28, "learning_rate": 3.2041871029329473e-06, "loss": 0.2522, "step": 28900 }, { "epoch": 4.29, "learning_rate": 3.1712696270449983e-06, "loss": 0.2542, "step": 28950 }, { "epoch": 4.3, "learning_rate": 3.1383521511570497e-06, "loss": 0.2844, "step": 29000 }, { "epoch": 4.3, "learning_rate": 3.1054346752691007e-06, "loss": 0.266, "step": 29050 }, { "epoch": 4.31, "learning_rate": 3.0725171993811513e-06, "loss": 0.2952, "step": 29100 }, { "epoch": 4.32, "learning_rate": 3.0395997234932028e-06, "loss": 0.2631, "step": 29150 }, { "epoch": 4.33, "learning_rate": 3.0066822476052538e-06, "loss": 0.2827, "step": 29200 }, { "epoch": 4.33, "learning_rate": 2.973764771717305e-06, "loss": 0.3, "step": 29250 }, { "epoch": 4.34, "learning_rate": 2.9408472958293562e-06, "loss": 0.2848, "step": 29300 }, { "epoch": 4.35, "learning_rate": 2.907929819941407e-06, "loss": 0.2997, "step": 29350 }, { "epoch": 4.35, "learning_rate": 2.8750123440534583e-06, "loss": 0.265, "step": 29400 }, { "epoch": 4.36, "learning_rate": 2.8420948681655093e-06, "loss": 0.2721, "step": 29450 }, { "epoch": 4.37, "learning_rate": 2.8091773922775607e-06, "loss": 0.2835, "step": 29500 }, { "epoch": 4.38, "learning_rate": 2.7762599163896113e-06, "loss": 0.2714, "step": 29550 }, { "epoch": 4.38, "learning_rate": 2.7433424405016623e-06, "loss": 0.2706, "step": 29600 }, { "epoch": 4.39, "learning_rate": 2.7104249646137138e-06, "loss": 0.2822, "step": 29650 }, { "epoch": 4.4, "learning_rate": 2.6775074887257648e-06, "loss": 0.2646, "step": 29700 }, { "epoch": 4.41, "learning_rate": 2.644590012837816e-06, "loss": 0.2709, "step": 29750 }, { "epoch": 4.41, "learning_rate": 2.611672536949867e-06, "loss": 0.2589, "step": 29800 }, { "epoch": 4.42, "learning_rate": 2.578755061061918e-06, "loss": 0.2949, "step": 29850 }, { "epoch": 4.43, "learning_rate": 2.5458375851739693e-06, "loss": 0.2912, "step": 29900 }, { "epoch": 4.44, "learning_rate": 2.51292010928602e-06, "loss": 0.279, "step": 29950 }, { "epoch": 4.44, "learning_rate": 2.48066098291583e-06, "loss": 0.259, "step": 30000 }, { "epoch": 4.44, "eval_loss": 0.12439576536417007, "eval_runtime": 93.9734, "eval_samples_per_second": 59.155, "eval_steps_per_second": 7.396, "eval_wer": 0.08736263212573508, "step": 30000 }, { "epoch": 4.45, "learning_rate": 2.447743507027881e-06, "loss": 0.2605, "step": 30050 }, { "epoch": 4.46, "learning_rate": 2.414826031139932e-06, "loss": 0.2597, "step": 30100 }, { "epoch": 4.47, "learning_rate": 2.3819085552519836e-06, "loss": 0.2725, "step": 30150 }, { "epoch": 4.47, "learning_rate": 2.3489910793640342e-06, "loss": 0.2677, "step": 30200 }, { "epoch": 4.48, "learning_rate": 2.3160736034760857e-06, "loss": 0.2731, "step": 30250 }, { "epoch": 4.49, "learning_rate": 2.2831561275881367e-06, "loss": 0.2554, "step": 30300 }, { "epoch": 4.5, "learning_rate": 2.2502386517001877e-06, "loss": 0.2856, "step": 30350 }, { "epoch": 4.5, "learning_rate": 2.217321175812239e-06, "loss": 0.2628, "step": 30400 }, { "epoch": 4.51, "learning_rate": 2.1844036999242897e-06, "loss": 0.275, "step": 30450 }, { "epoch": 4.52, "learning_rate": 2.151486224036341e-06, "loss": 0.2818, "step": 30500 }, { "epoch": 4.53, "learning_rate": 2.118568748148392e-06, "loss": 0.2712, "step": 30550 }, { "epoch": 4.53, "learning_rate": 2.085651272260443e-06, "loss": 0.3019, "step": 30600 }, { "epoch": 4.54, "learning_rate": 2.0527337963724942e-06, "loss": 0.2894, "step": 30650 }, { "epoch": 4.55, "learning_rate": 2.0198163204845452e-06, "loss": 0.254, "step": 30700 }, { "epoch": 4.55, "learning_rate": 1.9868988445965962e-06, "loss": 0.276, "step": 30750 }, { "epoch": 4.56, "learning_rate": 1.9539813687086477e-06, "loss": 0.2779, "step": 30800 }, { "epoch": 4.57, "learning_rate": 1.9210638928206987e-06, "loss": 0.26, "step": 30850 }, { "epoch": 4.58, "learning_rate": 1.88814641693275e-06, "loss": 0.2753, "step": 30900 }, { "epoch": 4.58, "learning_rate": 1.8552289410448007e-06, "loss": 0.2761, "step": 30950 }, { "epoch": 4.59, "learning_rate": 1.822311465156852e-06, "loss": 0.274, "step": 31000 }, { "epoch": 4.6, "learning_rate": 1.789393989268903e-06, "loss": 0.2962, "step": 31050 }, { "epoch": 4.61, "learning_rate": 1.7564765133809542e-06, "loss": 0.2814, "step": 31100 }, { "epoch": 4.61, "learning_rate": 1.7235590374930052e-06, "loss": 0.2748, "step": 31150 }, { "epoch": 4.62, "learning_rate": 1.6906415616050562e-06, "loss": 0.265, "step": 31200 }, { "epoch": 4.63, "learning_rate": 1.6577240857171072e-06, "loss": 0.261, "step": 31250 }, { "epoch": 4.64, "learning_rate": 1.6248066098291585e-06, "loss": 0.2743, "step": 31300 }, { "epoch": 4.64, "learning_rate": 1.5918891339412095e-06, "loss": 0.2733, "step": 31350 }, { "epoch": 4.65, "learning_rate": 1.5589716580532607e-06, "loss": 0.2682, "step": 31400 }, { "epoch": 4.66, "learning_rate": 1.5260541821653115e-06, "loss": 0.294, "step": 31450 }, { "epoch": 4.67, "learning_rate": 1.4931367062773627e-06, "loss": 0.2913, "step": 31500 }, { "epoch": 4.67, "learning_rate": 1.460219230389414e-06, "loss": 0.2827, "step": 31550 }, { "epoch": 4.68, "learning_rate": 1.427301754501465e-06, "loss": 0.2742, "step": 31600 }, { "epoch": 4.69, "learning_rate": 1.3943842786135162e-06, "loss": 0.2746, "step": 31650 }, { "epoch": 4.7, "learning_rate": 1.361466802725567e-06, "loss": 0.2819, "step": 31700 }, { "epoch": 4.7, "learning_rate": 1.3285493268376182e-06, "loss": 0.2571, "step": 31750 }, { "epoch": 4.71, "learning_rate": 1.2956318509496693e-06, "loss": 0.2589, "step": 31800 }, { "epoch": 4.72, "learning_rate": 1.2627143750617205e-06, "loss": 0.2766, "step": 31850 }, { "epoch": 4.73, "learning_rate": 1.2297968991737715e-06, "loss": 0.2678, "step": 31900 }, { "epoch": 4.73, "learning_rate": 1.1968794232858225e-06, "loss": 0.262, "step": 31950 }, { "epoch": 4.74, "learning_rate": 1.1639619473978735e-06, "loss": 0.2704, "step": 32000 }, { "epoch": 4.75, "learning_rate": 1.1310444715099248e-06, "loss": 0.2684, "step": 32050 }, { "epoch": 4.75, "learning_rate": 1.0981269956219758e-06, "loss": 0.2669, "step": 32100 }, { "epoch": 4.76, "learning_rate": 1.065209519734027e-06, "loss": 0.2673, "step": 32150 }, { "epoch": 4.77, "learning_rate": 1.032292043846078e-06, "loss": 0.2684, "step": 32200 }, { "epoch": 4.78, "learning_rate": 1.000032917475888e-06, "loss": 0.2592, "step": 32250 }, { "epoch": 4.78, "learning_rate": 9.671154415879391e-07, "loss": 0.2842, "step": 32300 }, { "epoch": 4.79, "learning_rate": 9.341979656999902e-07, "loss": 0.2798, "step": 32350 }, { "epoch": 4.8, "learning_rate": 9.012804898120413e-07, "loss": 0.2715, "step": 32400 }, { "epoch": 4.81, "learning_rate": 8.683630139240924e-07, "loss": 0.2621, "step": 32450 }, { "epoch": 4.81, "learning_rate": 8.354455380361434e-07, "loss": 0.2348, "step": 32500 }, { "epoch": 4.81, "eval_loss": 0.12447884678840637, "eval_runtime": 94.4113, "eval_samples_per_second": 58.881, "eval_steps_per_second": 7.361, "eval_wer": 0.08647623404721738, "step": 32500 }, { "epoch": 4.82, "learning_rate": 8.025280621481945e-07, "loss": 0.27, "step": 32550 }, { "epoch": 4.83, "learning_rate": 7.696105862602456e-07, "loss": 0.2432, "step": 32600 }, { "epoch": 4.84, "learning_rate": 7.366931103722967e-07, "loss": 0.2449, "step": 32650 }, { "epoch": 4.84, "learning_rate": 7.037756344843479e-07, "loss": 0.2593, "step": 32700 }, { "epoch": 4.85, "learning_rate": 6.708581585963988e-07, "loss": 0.2611, "step": 32750 }, { "epoch": 4.86, "learning_rate": 6.3794068270845e-07, "loss": 0.2775, "step": 32800 }, { "epoch": 4.87, "learning_rate": 6.050232068205011e-07, "loss": 0.2491, "step": 32850 }, { "epoch": 4.87, "learning_rate": 5.721057309325522e-07, "loss": 0.2632, "step": 32900 }, { "epoch": 4.88, "learning_rate": 5.391882550446032e-07, "loss": 0.2647, "step": 32950 }, { "epoch": 4.89, "learning_rate": 5.062707791566543e-07, "loss": 0.2353, "step": 33000 }, { "epoch": 4.9, "learning_rate": 4.733533032687054e-07, "loss": 0.2626, "step": 33050 }, { "epoch": 4.9, "learning_rate": 4.404358273807565e-07, "loss": 0.2872, "step": 33100 }, { "epoch": 4.91, "learning_rate": 4.0751835149280757e-07, "loss": 0.278, "step": 33150 }, { "epoch": 4.92, "learning_rate": 3.7460087560485863e-07, "loss": 0.2811, "step": 33200 }, { "epoch": 4.93, "learning_rate": 3.416833997169097e-07, "loss": 0.2887, "step": 33250 }, { "epoch": 4.93, "learning_rate": 3.087659238289608e-07, "loss": 0.307, "step": 33300 }, { "epoch": 4.94, "learning_rate": 2.758484479410119e-07, "loss": 0.294, "step": 33350 }, { "epoch": 4.95, "learning_rate": 2.4293097205306296e-07, "loss": 0.2594, "step": 33400 }, { "epoch": 4.95, "learning_rate": 2.1001349616511408e-07, "loss": 0.2641, "step": 33450 }, { "epoch": 4.96, "learning_rate": 1.7709602027716515e-07, "loss": 0.2671, "step": 33500 }, { "epoch": 4.97, "learning_rate": 1.4417854438921624e-07, "loss": 0.2642, "step": 33550 }, { "epoch": 4.98, "learning_rate": 1.1126106850126734e-07, "loss": 0.264, "step": 33600 }, { "epoch": 4.98, "learning_rate": 7.834359261331842e-08, "loss": 0.2958, "step": 33650 }, { "epoch": 4.99, "learning_rate": 4.5426116725369505e-08, "loss": 0.2434, "step": 33700 }, { "epoch": 5.0, "learning_rate": 1.2508640837420587e-08, "loss": 0.2606, "step": 33750 }, { "epoch": 5.0, "step": 33755, "total_flos": 6.405034708820275e+20, "train_loss": 0.7237436476221368, "train_runtime": 32222.0782, "train_samples_per_second": 41.901, "train_steps_per_second": 1.048 } ], "logging_steps": 50, "max_steps": 33755, "num_train_epochs": 5, "save_steps": 500, "total_flos": 6.405034708820275e+20, "trial_name": null, "trial_params": null }