{
"best_metric": 0.1262135922330097,
"best_model_checkpoint": "results3\\checkpoint-124000",
"epoch": 2.002992,
"eval_steps": 4000,
"global_step": 124000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0008,
"grad_norm": 12.33169937133789,
"learning_rate": 1.8800000000000002e-06,
"loss": 1.7644,
"step": 100
},
{
"epoch": 0.0016,
"grad_norm": 9.583354949951172,
"learning_rate": 3.88e-06,
"loss": 0.929,
"step": 200
},
{
"epoch": 0.0024,
"grad_norm": 12.06747817993164,
"learning_rate": 5.8800000000000005e-06,
"loss": 0.4652,
"step": 300
},
{
"epoch": 0.0032,
"grad_norm": 9.123275756835938,
"learning_rate": 7.88e-06,
"loss": 0.3351,
"step": 400
},
{
"epoch": 0.004,
"grad_norm": 15.981216430664062,
"learning_rate": 9.88e-06,
"loss": 0.4079,
"step": 500
},
{
"epoch": 0.0048,
"grad_norm": 20.674148559570312,
"learning_rate": 9.992449799196789e-06,
"loss": 0.3966,
"step": 600
},
{
"epoch": 0.0056,
"grad_norm": 11.142045974731445,
"learning_rate": 9.984417670682733e-06,
"loss": 0.3628,
"step": 700
},
{
"epoch": 0.0064,
"grad_norm": 6.716693878173828,
"learning_rate": 9.976385542168675e-06,
"loss": 0.3647,
"step": 800
},
{
"epoch": 0.0072,
"grad_norm": 15.127047538757324,
"learning_rate": 9.968353413654619e-06,
"loss": 0.4007,
"step": 900
},
{
"epoch": 0.008,
"grad_norm": 5.561180114746094,
"learning_rate": 9.960321285140563e-06,
"loss": 0.3748,
"step": 1000
},
{
"epoch": 0.0088,
"grad_norm": 11.087468147277832,
"learning_rate": 9.952289156626507e-06,
"loss": 0.3802,
"step": 1100
},
{
"epoch": 0.0096,
"grad_norm": 13.053077697753906,
"learning_rate": 9.94425702811245e-06,
"loss": 0.3555,
"step": 1200
},
{
"epoch": 0.0104,
"grad_norm": 8.72202205657959,
"learning_rate": 9.936224899598395e-06,
"loss": 0.3532,
"step": 1300
},
{
"epoch": 0.0112,
"grad_norm": 13.878268241882324,
"learning_rate": 9.928192771084338e-06,
"loss": 0.3571,
"step": 1400
},
{
"epoch": 0.012,
"grad_norm": 7.951385498046875,
"learning_rate": 9.920160642570282e-06,
"loss": 0.3683,
"step": 1500
},
{
"epoch": 0.0128,
"grad_norm": 7.835740566253662,
"learning_rate": 9.912128514056226e-06,
"loss": 0.3409,
"step": 1600
},
{
"epoch": 0.0136,
"grad_norm": 10.594538688659668,
"learning_rate": 9.904096385542169e-06,
"loss": 0.3455,
"step": 1700
},
{
"epoch": 0.0144,
"grad_norm": 8.736641883850098,
"learning_rate": 9.896064257028112e-06,
"loss": 0.3567,
"step": 1800
},
{
"epoch": 0.0152,
"grad_norm": 6.575601100921631,
"learning_rate": 9.888032128514056e-06,
"loss": 0.3606,
"step": 1900
},
{
"epoch": 0.016,
"grad_norm": 11.653617858886719,
"learning_rate": 9.88e-06,
"loss": 0.3683,
"step": 2000
},
{
"epoch": 0.0168,
"grad_norm": 12.467767715454102,
"learning_rate": 9.871967871485944e-06,
"loss": 0.3723,
"step": 2100
},
{
"epoch": 0.0176,
"grad_norm": 11.205801963806152,
"learning_rate": 9.863935742971888e-06,
"loss": 0.3561,
"step": 2200
},
{
"epoch": 0.0184,
"grad_norm": 10.407919883728027,
"learning_rate": 9.855903614457832e-06,
"loss": 0.3414,
"step": 2300
},
{
"epoch": 0.0192,
"grad_norm": 6.497184753417969,
"learning_rate": 9.847871485943776e-06,
"loss": 0.3439,
"step": 2400
},
{
"epoch": 0.02,
"grad_norm": 10.192763328552246,
"learning_rate": 9.83983935742972e-06,
"loss": 0.3334,
"step": 2500
},
{
"epoch": 0.0208,
"grad_norm": 10.233382225036621,
"learning_rate": 9.831807228915664e-06,
"loss": 0.3715,
"step": 2600
},
{
"epoch": 0.0216,
"grad_norm": 10.373730659484863,
"learning_rate": 9.823775100401608e-06,
"loss": 0.3557,
"step": 2700
},
{
"epoch": 0.0224,
"grad_norm": 9.90769100189209,
"learning_rate": 9.81574297188755e-06,
"loss": 0.345,
"step": 2800
},
{
"epoch": 0.0232,
"grad_norm": 12.89505386352539,
"learning_rate": 9.807710843373494e-06,
"loss": 0.3466,
"step": 2900
},
{
"epoch": 0.024,
"grad_norm": 11.755387306213379,
"learning_rate": 9.799678714859438e-06,
"loss": 0.3518,
"step": 3000
},
{
"epoch": 0.0248,
"grad_norm": 23.758481979370117,
"learning_rate": 9.791646586345382e-06,
"loss": 0.3367,
"step": 3100
},
{
"epoch": 0.0256,
"grad_norm": 8.041315078735352,
"learning_rate": 9.783614457831326e-06,
"loss": 0.3385,
"step": 3200
},
{
"epoch": 0.0264,
"grad_norm": 6.949316501617432,
"learning_rate": 9.77558232931727e-06,
"loss": 0.3557,
"step": 3300
},
{
"epoch": 0.0272,
"grad_norm": 20.79637908935547,
"learning_rate": 9.767550200803213e-06,
"loss": 0.3497,
"step": 3400
},
{
"epoch": 0.028,
"grad_norm": 10.800033569335938,
"learning_rate": 9.759518072289157e-06,
"loss": 0.351,
"step": 3500
},
{
"epoch": 0.0288,
"grad_norm": 7.772594451904297,
"learning_rate": 9.751485943775101e-06,
"loss": 0.3272,
"step": 3600
},
{
"epoch": 0.0296,
"grad_norm": 7.19946813583374,
"learning_rate": 9.743453815261045e-06,
"loss": 0.3216,
"step": 3700
},
{
"epoch": 0.0304,
"grad_norm": 11.165406227111816,
"learning_rate": 9.735421686746989e-06,
"loss": 0.3305,
"step": 3800
},
{
"epoch": 0.0312,
"grad_norm": 12.210046768188477,
"learning_rate": 9.727389558232933e-06,
"loss": 0.3524,
"step": 3900
},
{
"epoch": 0.032,
"grad_norm": 6.451345443725586,
"learning_rate": 9.719357429718877e-06,
"loss": 0.3394,
"step": 4000
},
{
"epoch": 0.032,
"eval_test1_cer": 0.09286074389424154,
"eval_test1_cer_norm": 0.07516507791677672,
"eval_test1_loss": 0.2566915452480316,
"eval_test1_runtime": 6436.6351,
"eval_test1_samples_per_second": 0.388,
"eval_test1_steps_per_second": 0.097,
"eval_test1_wer": 0.2368582174407417,
"eval_test1_wer_norm": 0.17320786697448787,
"step": 4000
},
{
"epoch": 0.032,
"eval_test2_cer": 0.2131304364057192,
"eval_test2_cer_norm": 0.18118124418964984,
"eval_test2_loss": 0.4192444682121277,
"eval_test2_runtime": 3419.6713,
"eval_test2_samples_per_second": 0.731,
"eval_test2_steps_per_second": 0.183,
"eval_test2_wer": 0.42229343099107347,
"eval_test2_wer_norm": 0.35425739170295667,
"step": 4000
},
{
"epoch": 0.0328,
"grad_norm": 43.34160232543945,
"learning_rate": 9.711325301204821e-06,
"loss": 0.3384,
"step": 4100
},
{
"epoch": 0.0336,
"grad_norm": 7.907329559326172,
"learning_rate": 9.703293172690765e-06,
"loss": 0.3674,
"step": 4200
},
{
"epoch": 0.0344,
"grad_norm": 6.71475076675415,
"learning_rate": 9.695261044176709e-06,
"loss": 0.3364,
"step": 4300
},
{
"epoch": 0.0352,
"grad_norm": 9.054850578308105,
"learning_rate": 9.687228915662651e-06,
"loss": 0.3417,
"step": 4400
},
{
"epoch": 0.036,
"grad_norm": 14.793953895568848,
"learning_rate": 9.679196787148595e-06,
"loss": 0.3286,
"step": 4500
},
{
"epoch": 0.0368,
"grad_norm": 5.203219413757324,
"learning_rate": 9.671164658634539e-06,
"loss": 0.3684,
"step": 4600
},
{
"epoch": 0.0376,
"grad_norm": 12.02910327911377,
"learning_rate": 9.663132530120483e-06,
"loss": 0.3393,
"step": 4700
},
{
"epoch": 0.0384,
"grad_norm": 10.032449722290039,
"learning_rate": 9.655100401606427e-06,
"loss": 0.3503,
"step": 4800
},
{
"epoch": 0.0392,
"grad_norm": 9.57917594909668,
"learning_rate": 9.64706827309237e-06,
"loss": 0.3239,
"step": 4900
},
{
"epoch": 0.04,
"grad_norm": 9.18687629699707,
"learning_rate": 9.639036144578314e-06,
"loss": 0.3407,
"step": 5000
},
{
"epoch": 0.0408,
"grad_norm": 9.08846378326416,
"learning_rate": 9.631004016064258e-06,
"loss": 0.3478,
"step": 5100
},
{
"epoch": 0.0416,
"grad_norm": 6.322196006774902,
"learning_rate": 9.622971887550202e-06,
"loss": 0.3294,
"step": 5200
},
{
"epoch": 0.0424,
"grad_norm": 11.071808815002441,
"learning_rate": 9.614939759036145e-06,
"loss": 0.3533,
"step": 5300
},
{
"epoch": 0.0432,
"grad_norm": 8.492830276489258,
"learning_rate": 9.606907630522088e-06,
"loss": 0.3297,
"step": 5400
},
{
"epoch": 0.044,
"grad_norm": 13.28835678100586,
"learning_rate": 9.598875502008032e-06,
"loss": 0.3228,
"step": 5500
},
{
"epoch": 0.0448,
"grad_norm": 9.030045509338379,
"learning_rate": 9.590843373493976e-06,
"loss": 0.3228,
"step": 5600
},
{
"epoch": 0.0456,
"grad_norm": 12.364481925964355,
"learning_rate": 9.58281124497992e-06,
"loss": 0.3406,
"step": 5700
},
{
"epoch": 0.0464,
"grad_norm": 6.908151149749756,
"learning_rate": 9.574779116465864e-06,
"loss": 0.3365,
"step": 5800
},
{
"epoch": 0.0472,
"grad_norm": 10.87695598602295,
"learning_rate": 9.566746987951808e-06,
"loss": 0.3332,
"step": 5900
},
{
"epoch": 0.048,
"grad_norm": 9.280797004699707,
"learning_rate": 9.558714859437752e-06,
"loss": 0.3414,
"step": 6000
},
{
"epoch": 0.0488,
"grad_norm": 13.637160301208496,
"learning_rate": 9.550682730923696e-06,
"loss": 0.3275,
"step": 6100
},
{
"epoch": 0.0496,
"grad_norm": 10.094959259033203,
"learning_rate": 9.542650602409638e-06,
"loss": 0.3152,
"step": 6200
},
{
"epoch": 0.0504,
"grad_norm": 6.261775016784668,
"learning_rate": 9.534618473895582e-06,
"loss": 0.3236,
"step": 6300
},
{
"epoch": 0.0512,
"grad_norm": 7.221704006195068,
"learning_rate": 9.526586345381526e-06,
"loss": 0.3461,
"step": 6400
},
{
"epoch": 0.052,
"grad_norm": 10.923733711242676,
"learning_rate": 9.51855421686747e-06,
"loss": 0.3249,
"step": 6500
},
{
"epoch": 0.0528,
"grad_norm": 5.432595729827881,
"learning_rate": 9.510522088353414e-06,
"loss": 0.3556,
"step": 6600
},
{
"epoch": 0.0536,
"grad_norm": 15.021200180053711,
"learning_rate": 9.502489959839358e-06,
"loss": 0.3208,
"step": 6700
},
{
"epoch": 0.0544,
"grad_norm": 6.641680717468262,
"learning_rate": 9.494457831325302e-06,
"loss": 0.3565,
"step": 6800
},
{
"epoch": 0.0552,
"grad_norm": 15.715243339538574,
"learning_rate": 9.486425702811246e-06,
"loss": 0.3123,
"step": 6900
},
{
"epoch": 0.056,
"grad_norm": 9.304129600524902,
"learning_rate": 9.47839357429719e-06,
"loss": 0.312,
"step": 7000
},
{
"epoch": 0.0568,
"grad_norm": 10.780037879943848,
"learning_rate": 9.470361445783133e-06,
"loss": 0.3445,
"step": 7100
},
{
"epoch": 0.0576,
"grad_norm": 11.77072811126709,
"learning_rate": 9.462329317269077e-06,
"loss": 0.3293,
"step": 7200
},
{
"epoch": 0.0584,
"grad_norm": 11.29255485534668,
"learning_rate": 9.454297188755021e-06,
"loss": 0.3045,
"step": 7300
},
{
"epoch": 0.0592,
"grad_norm": 9.00106143951416,
"learning_rate": 9.446265060240965e-06,
"loss": 0.3132,
"step": 7400
},
{
"epoch": 0.06,
"grad_norm": 8.445959091186523,
"learning_rate": 9.438232931726909e-06,
"loss": 0.3327,
"step": 7500
},
{
"epoch": 0.0608,
"grad_norm": 13.996682167053223,
"learning_rate": 9.430200803212853e-06,
"loss": 0.3496,
"step": 7600
},
{
"epoch": 0.0616,
"grad_norm": 6.321784019470215,
"learning_rate": 9.422168674698797e-06,
"loss": 0.2928,
"step": 7700
},
{
"epoch": 0.0624,
"grad_norm": 9.260029792785645,
"learning_rate": 9.414136546184741e-06,
"loss": 0.2978,
"step": 7800
},
{
"epoch": 0.0632,
"grad_norm": 14.37282657623291,
"learning_rate": 9.406104417670685e-06,
"loss": 0.3177,
"step": 7900
},
{
"epoch": 0.064,
"grad_norm": 13.508281707763672,
"learning_rate": 9.398072289156627e-06,
"loss": 0.3357,
"step": 8000
},
{
"epoch": 0.064,
"eval_test1_cer": 0.07946822018074538,
"eval_test1_cer_norm": 0.05569188657046126,
"eval_test1_loss": 0.2466106414794922,
"eval_test1_runtime": 3431.7176,
"eval_test1_samples_per_second": 0.728,
"eval_test1_steps_per_second": 0.182,
"eval_test1_wer": 0.21889851016064607,
"eval_test1_wer_norm": 0.15733948975715245,
"step": 8000
},
{
"epoch": 0.064,
"eval_test2_cer": 0.21952346287378205,
"eval_test2_cer_norm": 0.16622927641772545,
"eval_test2_loss": 0.410643607378006,
"eval_test2_runtime": 3658.9711,
"eval_test2_samples_per_second": 0.683,
"eval_test2_steps_per_second": 0.171,
"eval_test2_wer": 0.39210917830167086,
"eval_test2_wer_norm": 0.32348727939491173,
"step": 8000
},
{
"epoch": 0.0648,
"grad_norm": 7.349660873413086,
"learning_rate": 9.390040160642571e-06,
"loss": 0.3311,
"step": 8100
},
{
"epoch": 0.0656,
"grad_norm": 6.557274341583252,
"learning_rate": 9.382008032128515e-06,
"loss": 0.318,
"step": 8200
},
{
"epoch": 0.0664,
"grad_norm": 20.285615921020508,
"learning_rate": 9.373975903614459e-06,
"loss": 0.3116,
"step": 8300
},
{
"epoch": 0.0672,
"grad_norm": 7.684230327606201,
"learning_rate": 9.365943775100403e-06,
"loss": 0.3232,
"step": 8400
},
{
"epoch": 0.068,
"grad_norm": 8.374768257141113,
"learning_rate": 9.357911646586347e-06,
"loss": 0.3328,
"step": 8500
},
{
"epoch": 0.0688,
"grad_norm": 6.924304008483887,
"learning_rate": 9.34987951807229e-06,
"loss": 0.2923,
"step": 8600
},
{
"epoch": 0.0696,
"grad_norm": 5.943380832672119,
"learning_rate": 9.341847389558234e-06,
"loss": 0.3167,
"step": 8700
},
{
"epoch": 0.0704,
"grad_norm": 24.188907623291016,
"learning_rate": 9.333815261044178e-06,
"loss": 0.3521,
"step": 8800
},
{
"epoch": 0.0712,
"grad_norm": 13.702215194702148,
"learning_rate": 9.325783132530122e-06,
"loss": 0.2937,
"step": 8900
},
{
"epoch": 0.072,
"grad_norm": 10.463766098022461,
"learning_rate": 9.317751004016065e-06,
"loss": 0.3317,
"step": 9000
},
{
"epoch": 0.0728,
"grad_norm": 5.035882472991943,
"learning_rate": 9.309718875502008e-06,
"loss": 0.3048,
"step": 9100
},
{
"epoch": 0.0736,
"grad_norm": 11.715450286865234,
"learning_rate": 9.301686746987952e-06,
"loss": 0.3287,
"step": 9200
},
{
"epoch": 0.0744,
"grad_norm": 13.449250221252441,
"learning_rate": 9.293654618473896e-06,
"loss": 0.3345,
"step": 9300
},
{
"epoch": 0.0752,
"grad_norm": 8.16373348236084,
"learning_rate": 9.28562248995984e-06,
"loss": 0.3233,
"step": 9400
},
{
"epoch": 0.076,
"grad_norm": 6.236880302429199,
"learning_rate": 9.277590361445784e-06,
"loss": 0.3131,
"step": 9500
},
{
"epoch": 0.0768,
"grad_norm": 12.164971351623535,
"learning_rate": 9.269558232931728e-06,
"loss": 0.3281,
"step": 9600
},
{
"epoch": 0.0776,
"grad_norm": 15.774728775024414,
"learning_rate": 9.261526104417672e-06,
"loss": 0.315,
"step": 9700
},
{
"epoch": 0.0784,
"grad_norm": 16.81951332092285,
"learning_rate": 9.253493975903616e-06,
"loss": 0.3491,
"step": 9800
},
{
"epoch": 0.0792,
"grad_norm": 35.960697174072266,
"learning_rate": 9.245461847389558e-06,
"loss": 0.3258,
"step": 9900
},
{
"epoch": 0.08,
"grad_norm": 9.389575958251953,
"learning_rate": 9.237429718875502e-06,
"loss": 0.3082,
"step": 10000
},
{
"epoch": 0.0808,
"grad_norm": 5.77083683013916,
"learning_rate": 9.229477911646586e-06,
"loss": 0.316,
"step": 10100
},
{
"epoch": 0.0816,
"grad_norm": 7.730663776397705,
"learning_rate": 9.22144578313253e-06,
"loss": 0.3538,
"step": 10200
},
{
"epoch": 0.0824,
"grad_norm": 2.672037363052368,
"learning_rate": 9.213413654618474e-06,
"loss": 0.3089,
"step": 10300
},
{
"epoch": 0.0832,
"grad_norm": 10.987016677856445,
"learning_rate": 9.205381526104418e-06,
"loss": 0.3032,
"step": 10400
},
{
"epoch": 0.084,
"grad_norm": 5.871443271636963,
"learning_rate": 9.197349397590362e-06,
"loss": 0.3423,
"step": 10500
},
{
"epoch": 0.0848,
"grad_norm": 8.393383026123047,
"learning_rate": 9.189317269076306e-06,
"loss": 0.2872,
"step": 10600
},
{
"epoch": 0.0856,
"grad_norm": 8.698946952819824,
"learning_rate": 9.18128514056225e-06,
"loss": 0.3028,
"step": 10700
},
{
"epoch": 0.0864,
"grad_norm": 2.9550344944000244,
"learning_rate": 9.173253012048194e-06,
"loss": 0.3262,
"step": 10800
},
{
"epoch": 0.0872,
"grad_norm": 6.455613136291504,
"learning_rate": 9.165220883534138e-06,
"loss": 0.3273,
"step": 10900
},
{
"epoch": 0.088,
"grad_norm": 11.550117492675781,
"learning_rate": 9.157269076305222e-06,
"loss": 0.3307,
"step": 11000
},
{
"epoch": 0.0888,
"grad_norm": 9.657747268676758,
"learning_rate": 9.149236947791166e-06,
"loss": 0.3227,
"step": 11100
},
{
"epoch": 0.0896,
"grad_norm": 9.803390502929688,
"learning_rate": 9.14120481927711e-06,
"loss": 0.3102,
"step": 11200
},
{
"epoch": 0.0904,
"grad_norm": 12.699715614318848,
"learning_rate": 9.133172690763052e-06,
"loss": 0.3353,
"step": 11300
},
{
"epoch": 0.0912,
"grad_norm": 9.973093032836914,
"learning_rate": 9.125140562248996e-06,
"loss": 0.3053,
"step": 11400
},
{
"epoch": 0.092,
"grad_norm": 12.822247505187988,
"learning_rate": 9.11710843373494e-06,
"loss": 0.3088,
"step": 11500
},
{
"epoch": 0.0928,
"grad_norm": 13.785049438476562,
"learning_rate": 9.109076305220884e-06,
"loss": 0.3004,
"step": 11600
},
{
"epoch": 0.0936,
"grad_norm": 11.756546020507812,
"learning_rate": 9.101044176706828e-06,
"loss": 0.304,
"step": 11700
},
{
"epoch": 0.0944,
"grad_norm": 12.297760009765625,
"learning_rate": 9.093012048192772e-06,
"loss": 0.3029,
"step": 11800
},
{
"epoch": 0.0952,
"grad_norm": 7.838765621185303,
"learning_rate": 9.084979919678716e-06,
"loss": 0.3046,
"step": 11900
},
{
"epoch": 0.096,
"grad_norm": 6.533823013305664,
"learning_rate": 9.07694779116466e-06,
"loss": 0.3065,
"step": 12000
},
{
"epoch": 0.096,
"eval_test1_cer": 0.10149656434386436,
"eval_test1_cer_norm": 0.06880207457920139,
"eval_test1_loss": 0.23942527174949646,
"eval_test1_runtime": 3496.8354,
"eval_test1_samples_per_second": 0.715,
"eval_test1_steps_per_second": 0.179,
"eval_test1_wer": 0.23831598588880143,
"eval_test1_wer_norm": 0.1713375610041205,
"step": 12000
},
{
"epoch": 0.096,
"eval_test2_cer": 0.2408444394669056,
"eval_test2_cer_norm": 0.178324488689185,
"eval_test2_loss": 0.3968297839164734,
"eval_test2_runtime": 3795.2628,
"eval_test2_samples_per_second": 0.659,
"eval_test2_steps_per_second": 0.165,
"eval_test2_wer": 0.4381437399862669,
"eval_test2_wer_norm": 0.36356864542745815,
"step": 12000
},
{
"epoch": 0.0968,
"grad_norm": 6.5714826583862305,
"learning_rate": 9.068915662650603e-06,
"loss": 0.3177,
"step": 12100
},
{
"epoch": 0.0976,
"grad_norm": 10.972187995910645,
"learning_rate": 9.060883534136546e-06,
"loss": 0.3032,
"step": 12200
},
{
"epoch": 0.0984,
"grad_norm": 12.001550674438477,
"learning_rate": 9.05285140562249e-06,
"loss": 0.3009,
"step": 12300
},
{
"epoch": 0.0992,
"grad_norm": 10.76689338684082,
"learning_rate": 9.044819277108433e-06,
"loss": 0.304,
"step": 12400
},
{
"epoch": 0.1,
"grad_norm": 6.645873069763184,
"learning_rate": 9.036787148594377e-06,
"loss": 0.2801,
"step": 12500
},
{
"epoch": 0.1008,
"grad_norm": 7.068800926208496,
"learning_rate": 9.028755020080321e-06,
"loss": 0.3198,
"step": 12600
},
{
"epoch": 0.1016,
"grad_norm": 7.24527645111084,
"learning_rate": 9.020722891566265e-06,
"loss": 0.3236,
"step": 12700
},
{
"epoch": 0.1024,
"grad_norm": 8.540508270263672,
"learning_rate": 9.01269076305221e-06,
"loss": 0.3132,
"step": 12800
},
{
"epoch": 0.1032,
"grad_norm": 10.710851669311523,
"learning_rate": 9.004658634538153e-06,
"loss": 0.3034,
"step": 12900
},
{
"epoch": 0.104,
"grad_norm": 7.53853702545166,
"learning_rate": 8.996626506024097e-06,
"loss": 0.2957,
"step": 13000
},
{
"epoch": 0.1048,
"grad_norm": 8.962101936340332,
"learning_rate": 8.988594377510041e-06,
"loss": 0.312,
"step": 13100
},
{
"epoch": 0.1056,
"grad_norm": 7.6881561279296875,
"learning_rate": 8.980642570281125e-06,
"loss": 0.3304,
"step": 13200
},
{
"epoch": 0.1064,
"grad_norm": 3.7053704261779785,
"learning_rate": 8.97261044176707e-06,
"loss": 0.2913,
"step": 13300
},
{
"epoch": 0.1072,
"grad_norm": 5.796387195587158,
"learning_rate": 8.964578313253013e-06,
"loss": 0.3162,
"step": 13400
},
{
"epoch": 0.108,
"grad_norm": 14.237732887268066,
"learning_rate": 8.956546184738957e-06,
"loss": 0.3096,
"step": 13500
},
{
"epoch": 0.1088,
"grad_norm": 9.336588859558105,
"learning_rate": 8.948514056224901e-06,
"loss": 0.3018,
"step": 13600
},
{
"epoch": 0.1096,
"grad_norm": 16.10430335998535,
"learning_rate": 8.940481927710845e-06,
"loss": 0.3149,
"step": 13700
},
{
"epoch": 0.1104,
"grad_norm": 70.292236328125,
"learning_rate": 8.932449799196787e-06,
"loss": 0.2721,
"step": 13800
},
{
"epoch": 0.1112,
"grad_norm": 5.2695770263671875,
"learning_rate": 8.924417670682731e-06,
"loss": 0.294,
"step": 13900
},
{
"epoch": 0.112,
"grad_norm": 7.514453411102295,
"learning_rate": 8.916385542168675e-06,
"loss": 0.3363,
"step": 14000
},
{
"epoch": 0.1128,
"grad_norm": 6.603350639343262,
"learning_rate": 8.908353413654619e-06,
"loss": 0.3214,
"step": 14100
},
{
"epoch": 0.1136,
"grad_norm": 12.648730278015137,
"learning_rate": 8.900321285140563e-06,
"loss": 0.313,
"step": 14200
},
{
"epoch": 0.1144,
"grad_norm": 19.52114486694336,
"learning_rate": 8.892289156626507e-06,
"loss": 0.2778,
"step": 14300
},
{
"epoch": 0.1152,
"grad_norm": 9.065464973449707,
"learning_rate": 8.88425702811245e-06,
"loss": 0.3182,
"step": 14400
},
{
"epoch": 0.116,
"grad_norm": 6.192761421203613,
"learning_rate": 8.876224899598395e-06,
"loss": 0.3268,
"step": 14500
},
{
"epoch": 0.1168,
"grad_norm": 6.336472034454346,
"learning_rate": 8.868192771084338e-06,
"loss": 0.3125,
"step": 14600
},
{
"epoch": 0.1176,
"grad_norm": 7.2869343757629395,
"learning_rate": 8.86016064257028e-06,
"loss": 0.2938,
"step": 14700
},
{
"epoch": 0.1184,
"grad_norm": 8.914804458618164,
"learning_rate": 8.852128514056225e-06,
"loss": 0.2831,
"step": 14800
},
{
"epoch": 0.1192,
"grad_norm": 8.745502471923828,
"learning_rate": 8.844096385542168e-06,
"loss": 0.306,
"step": 14900
},
{
"epoch": 0.12,
"grad_norm": 8.596148490905762,
"learning_rate": 8.836064257028112e-06,
"loss": 0.2907,
"step": 15000
},
{
"epoch": 0.1208,
"grad_norm": 6.889795303344727,
"learning_rate": 8.828032128514056e-06,
"loss": 0.2806,
"step": 15100
},
{
"epoch": 0.1216,
"grad_norm": 9.58425521850586,
"learning_rate": 8.82e-06,
"loss": 0.2853,
"step": 15200
},
{
"epoch": 0.1224,
"grad_norm": 5.967700004577637,
"learning_rate": 8.811967871485944e-06,
"loss": 0.3109,
"step": 15300
},
{
"epoch": 0.1232,
"grad_norm": 10.997330665588379,
"learning_rate": 8.80401606425703e-06,
"loss": 0.3129,
"step": 15400
},
{
"epoch": 0.124,
"grad_norm": 8.48721981048584,
"learning_rate": 8.795983935742972e-06,
"loss": 0.3227,
"step": 15500
},
{
"epoch": 0.1248,
"grad_norm": 9.547022819519043,
"learning_rate": 8.787951807228916e-06,
"loss": 0.282,
"step": 15600
},
{
"epoch": 0.1256,
"grad_norm": 12.319640159606934,
"learning_rate": 8.77991967871486e-06,
"loss": 0.3019,
"step": 15700
},
{
"epoch": 0.1264,
"grad_norm": 9.99170970916748,
"learning_rate": 8.771887550200804e-06,
"loss": 0.2872,
"step": 15800
},
{
"epoch": 0.1272,
"grad_norm": 9.307220458984375,
"learning_rate": 8.763855421686748e-06,
"loss": 0.2955,
"step": 15900
},
{
"epoch": 0.128,
"grad_norm": 6.70652961730957,
"learning_rate": 8.755823293172692e-06,
"loss": 0.2813,
"step": 16000
},
{
"epoch": 0.128,
"eval_test1_cer": 0.11614478303084622,
"eval_test1_cer_norm": 0.08919730112613153,
"eval_test1_loss": 0.2319698929786682,
"eval_test1_runtime": 2494.875,
"eval_test1_samples_per_second": 1.002,
"eval_test1_steps_per_second": 0.251,
"eval_test1_wer": 0.2696580075220852,
"eval_test1_wer_norm": 0.2076331862415617,
"step": 16000
},
{
"epoch": 0.128,
"eval_test2_cer": 0.25404113935864414,
"eval_test2_cer_norm": 0.1936541292221878,
"eval_test2_loss": 0.38276800513267517,
"eval_test2_runtime": 2638.5507,
"eval_test2_samples_per_second": 0.947,
"eval_test2_steps_per_second": 0.237,
"eval_test2_wer": 0.46260585946440835,
"eval_test2_wer_norm": 0.3951409580563832,
"step": 16000
},
{
"epoch": 0.1288,
"grad_norm": 9.956653594970703,
"learning_rate": 8.747791164658636e-06,
"loss": 0.2785,
"step": 16100
},
{
"epoch": 0.1296,
"grad_norm": 10.152132987976074,
"learning_rate": 8.73975903614458e-06,
"loss": 0.2918,
"step": 16200
},
{
"epoch": 0.1304,
"grad_norm": 10.379541397094727,
"learning_rate": 8.731726907630524e-06,
"loss": 0.2879,
"step": 16300
},
{
"epoch": 0.1312,
"grad_norm": 5.655648708343506,
"learning_rate": 8.723694779116466e-06,
"loss": 0.2935,
"step": 16400
},
{
"epoch": 0.132,
"grad_norm": 9.877130508422852,
"learning_rate": 8.71566265060241e-06,
"loss": 0.3331,
"step": 16500
},
{
"epoch": 0.1328,
"grad_norm": 8.812068939208984,
"learning_rate": 8.707630522088354e-06,
"loss": 0.2902,
"step": 16600
},
{
"epoch": 0.1336,
"grad_norm": 8.972885131835938,
"learning_rate": 8.699598393574298e-06,
"loss": 0.3017,
"step": 16700
},
{
"epoch": 0.1344,
"grad_norm": 6.562330722808838,
"learning_rate": 8.691566265060242e-06,
"loss": 0.2866,
"step": 16800
},
{
"epoch": 0.1352,
"grad_norm": 10.704404830932617,
"learning_rate": 8.683534136546186e-06,
"loss": 0.3078,
"step": 16900
},
{
"epoch": 0.136,
"grad_norm": 9.468811988830566,
"learning_rate": 8.67550200803213e-06,
"loss": 0.2906,
"step": 17000
},
{
"epoch": 0.1368,
"grad_norm": 5.418923854827881,
"learning_rate": 8.667469879518073e-06,
"loss": 0.2918,
"step": 17100
},
{
"epoch": 0.1376,
"grad_norm": 10.071795463562012,
"learning_rate": 8.659437751004017e-06,
"loss": 0.3149,
"step": 17200
},
{
"epoch": 0.1384,
"grad_norm": 10.977407455444336,
"learning_rate": 8.65140562248996e-06,
"loss": 0.3091,
"step": 17300
},
{
"epoch": 0.1392,
"grad_norm": 7.745774745941162,
"learning_rate": 8.643373493975904e-06,
"loss": 0.2843,
"step": 17400
},
{
"epoch": 0.14,
"grad_norm": 5.645944595336914,
"learning_rate": 8.635341365461847e-06,
"loss": 0.2824,
"step": 17500
},
{
"epoch": 0.1408,
"grad_norm": 11.466683387756348,
"learning_rate": 8.627309236947791e-06,
"loss": 0.3228,
"step": 17600
},
{
"epoch": 0.1416,
"grad_norm": 7.580329418182373,
"learning_rate": 8.619277108433735e-06,
"loss": 0.2967,
"step": 17700
},
{
"epoch": 0.1424,
"grad_norm": 11.873896598815918,
"learning_rate": 8.61124497991968e-06,
"loss": 0.3073,
"step": 17800
},
{
"epoch": 0.1432,
"grad_norm": 6.72385835647583,
"learning_rate": 8.603212851405623e-06,
"loss": 0.3105,
"step": 17900
},
{
"epoch": 0.144,
"grad_norm": 11.72990608215332,
"learning_rate": 8.595180722891567e-06,
"loss": 0.3141,
"step": 18000
},
{
"epoch": 0.1448,
"grad_norm": 16.776378631591797,
"learning_rate": 8.587148594377511e-06,
"loss": 0.2964,
"step": 18100
},
{
"epoch": 0.1456,
"grad_norm": 9.208036422729492,
"learning_rate": 8.579116465863455e-06,
"loss": 0.2818,
"step": 18200
},
{
"epoch": 0.1464,
"grad_norm": 10.971381187438965,
"learning_rate": 8.571084337349397e-06,
"loss": 0.2694,
"step": 18300
},
{
"epoch": 0.1472,
"grad_norm": 10.638858795166016,
"learning_rate": 8.563052208835341e-06,
"loss": 0.2927,
"step": 18400
},
{
"epoch": 0.148,
"grad_norm": 6.251519203186035,
"learning_rate": 8.555020080321285e-06,
"loss": 0.2835,
"step": 18500
},
{
"epoch": 0.1488,
"grad_norm": 10.34622859954834,
"learning_rate": 8.546987951807229e-06,
"loss": 0.298,
"step": 18600
},
{
"epoch": 0.1496,
"grad_norm": 9.303108215332031,
"learning_rate": 8.538955823293173e-06,
"loss": 0.2856,
"step": 18700
},
{
"epoch": 0.1504,
"grad_norm": 8.3678617477417,
"learning_rate": 8.530923694779117e-06,
"loss": 0.2782,
"step": 18800
},
{
"epoch": 0.1512,
"grad_norm": 11.090498924255371,
"learning_rate": 8.52289156626506e-06,
"loss": 0.2863,
"step": 18900
},
{
"epoch": 0.152,
"grad_norm": 5.836663722991943,
"learning_rate": 8.514859437751005e-06,
"loss": 0.2693,
"step": 19000
},
{
"epoch": 0.1528,
"grad_norm": 6.694278717041016,
"learning_rate": 8.506827309236948e-06,
"loss": 0.2817,
"step": 19100
},
{
"epoch": 0.1536,
"grad_norm": 5.725604057312012,
"learning_rate": 8.498795180722892e-06,
"loss": 0.265,
"step": 19200
},
{
"epoch": 0.1544,
"grad_norm": 8.070332527160645,
"learning_rate": 8.490763052208836e-06,
"loss": 0.302,
"step": 19300
},
{
"epoch": 0.1552,
"grad_norm": 7.241143226623535,
"learning_rate": 8.48281124497992e-06,
"loss": 0.2898,
"step": 19400
},
{
"epoch": 0.156,
"grad_norm": 8.699810981750488,
"learning_rate": 8.474779116465865e-06,
"loss": 0.2868,
"step": 19500
},
{
"epoch": 0.1568,
"grad_norm": 4.879644870758057,
"learning_rate": 8.466746987951808e-06,
"loss": 0.2917,
"step": 19600
},
{
"epoch": 0.1576,
"grad_norm": 6.502651691436768,
"learning_rate": 8.458714859437752e-06,
"loss": 0.2888,
"step": 19700
},
{
"epoch": 0.1584,
"grad_norm": 5.7493791580200195,
"learning_rate": 8.450682730923695e-06,
"loss": 0.2936,
"step": 19800
},
{
"epoch": 0.1592,
"grad_norm": 10.4688720703125,
"learning_rate": 8.442650602409639e-06,
"loss": 0.2806,
"step": 19900
},
{
"epoch": 0.16,
"grad_norm": 7.386699676513672,
"learning_rate": 8.434618473895582e-06,
"loss": 0.3055,
"step": 20000
},
{
"epoch": 0.16,
"eval_test1_cer": 0.09210452610351781,
"eval_test1_cer_norm": 0.07215885898143924,
"eval_test1_loss": 0.22505834698677063,
"eval_test1_runtime": 2739.9634,
"eval_test1_samples_per_second": 0.912,
"eval_test1_steps_per_second": 0.228,
"eval_test1_wer": 0.23679990670281933,
"eval_test1_wer_norm": 0.17437680820596743,
"step": 20000
},
{
"epoch": 0.16,
"eval_test2_cer": 0.15537387538731473,
"eval_test2_cer_norm": 0.12383792996591261,
"eval_test2_loss": 0.3772575259208679,
"eval_test2_runtime": 3715.4173,
"eval_test2_samples_per_second": 0.673,
"eval_test2_steps_per_second": 0.168,
"eval_test2_wer": 0.31795033188372623,
"eval_test2_wer_norm": 0.24954159981663993,
"step": 20000
},
{
"epoch": 0.1608,
"grad_norm": 9.69093132019043,
"learning_rate": 8.426586345381526e-06,
"loss": 0.2777,
"step": 20100
},
{
"epoch": 0.1616,
"grad_norm": 8.120789527893066,
"learning_rate": 8.41855421686747e-06,
"loss": 0.2859,
"step": 20200
},
{
"epoch": 0.1624,
"grad_norm": 7.084224700927734,
"learning_rate": 8.410522088353414e-06,
"loss": 0.3035,
"step": 20300
},
{
"epoch": 0.1632,
"grad_norm": 10.893829345703125,
"learning_rate": 8.402489959839358e-06,
"loss": 0.2792,
"step": 20400
},
{
"epoch": 0.164,
"grad_norm": 36.892173767089844,
"learning_rate": 8.394457831325302e-06,
"loss": 0.2835,
"step": 20500
},
{
"epoch": 0.1648,
"grad_norm": 8.867621421813965,
"learning_rate": 8.386425702811246e-06,
"loss": 0.2752,
"step": 20600
},
{
"epoch": 0.1656,
"grad_norm": 5.385016918182373,
"learning_rate": 8.378393574297188e-06,
"loss": 0.2763,
"step": 20700
},
{
"epoch": 0.1664,
"grad_norm": 6.431478023529053,
"learning_rate": 8.370361445783132e-06,
"loss": 0.2874,
"step": 20800
},
{
"epoch": 0.1672,
"grad_norm": 7.890101432800293,
"learning_rate": 8.362329317269076e-06,
"loss": 0.2737,
"step": 20900
},
{
"epoch": 0.168,
"grad_norm": 5.053267955780029,
"learning_rate": 8.35429718875502e-06,
"loss": 0.2885,
"step": 21000
},
{
"epoch": 0.1688,
"grad_norm": 7.141018390655518,
"learning_rate": 8.346265060240964e-06,
"loss": 0.2735,
"step": 21100
},
{
"epoch": 0.1696,
"grad_norm": 12.165709495544434,
"learning_rate": 8.338232931726908e-06,
"loss": 0.3086,
"step": 21200
},
{
"epoch": 0.1704,
"grad_norm": 5.507659912109375,
"learning_rate": 8.330200803212852e-06,
"loss": 0.2773,
"step": 21300
},
{
"epoch": 0.1712,
"grad_norm": 6.527307987213135,
"learning_rate": 8.322248995983936e-06,
"loss": 0.2584,
"step": 21400
},
{
"epoch": 0.172,
"grad_norm": 8.494378089904785,
"learning_rate": 8.31421686746988e-06,
"loss": 0.3005,
"step": 21500
},
{
"epoch": 0.1728,
"grad_norm": 6.213225364685059,
"learning_rate": 8.306184738955824e-06,
"loss": 0.2708,
"step": 21600
},
{
"epoch": 0.1736,
"grad_norm": 10.444897651672363,
"learning_rate": 8.298152610441768e-06,
"loss": 0.2817,
"step": 21700
},
{
"epoch": 0.1744,
"grad_norm": 6.7094950675964355,
"learning_rate": 8.290120481927712e-06,
"loss": 0.2555,
"step": 21800
},
{
"epoch": 0.1752,
"grad_norm": 7.899540901184082,
"learning_rate": 8.282088353413656e-06,
"loss": 0.2902,
"step": 21900
},
{
"epoch": 0.176,
"grad_norm": 7.049533367156982,
"learning_rate": 8.2740562248996e-06,
"loss": 0.2648,
"step": 22000
},
{
"epoch": 0.1768,
"grad_norm": 7.662154674530029,
"learning_rate": 8.266024096385543e-06,
"loss": 0.2984,
"step": 22100
},
{
"epoch": 0.1776,
"grad_norm": 5.362339019775391,
"learning_rate": 8.257991967871487e-06,
"loss": 0.2533,
"step": 22200
},
{
"epoch": 0.1784,
"grad_norm": 11.844931602478027,
"learning_rate": 8.249959839357431e-06,
"loss": 0.2967,
"step": 22300
},
{
"epoch": 0.1792,
"grad_norm": 8.356042861938477,
"learning_rate": 8.241927710843374e-06,
"loss": 0.2817,
"step": 22400
},
{
"epoch": 0.18,
"grad_norm": 6.400118827819824,
"learning_rate": 8.233895582329317e-06,
"loss": 0.2975,
"step": 22500
},
{
"epoch": 0.1808,
"grad_norm": 7.8221611976623535,
"learning_rate": 8.225863453815261e-06,
"loss": 0.2928,
"step": 22600
},
{
"epoch": 0.1816,
"grad_norm": 9.997267723083496,
"learning_rate": 8.217831325301205e-06,
"loss": 0.2846,
"step": 22700
},
{
"epoch": 0.1824,
"grad_norm": 3.741245746612549,
"learning_rate": 8.20979919678715e-06,
"loss": 0.2712,
"step": 22800
},
{
"epoch": 0.1832,
"grad_norm": 6.534188270568848,
"learning_rate": 8.201767068273093e-06,
"loss": 0.2879,
"step": 22900
},
{
"epoch": 0.184,
"grad_norm": 5.730849266052246,
"learning_rate": 8.193734939759037e-06,
"loss": 0.2722,
"step": 23000
},
{
"epoch": 0.1848,
"grad_norm": 4.018589973449707,
"learning_rate": 8.185702811244981e-06,
"loss": 0.2979,
"step": 23100
},
{
"epoch": 0.1856,
"grad_norm": 7.68761682510376,
"learning_rate": 8.177670682730925e-06,
"loss": 0.2919,
"step": 23200
},
{
"epoch": 0.1864,
"grad_norm": 9.94321060180664,
"learning_rate": 8.169638554216867e-06,
"loss": 0.2731,
"step": 23300
},
{
"epoch": 0.1872,
"grad_norm": 2.9645307064056396,
"learning_rate": 8.161686746987953e-06,
"loss": 0.2698,
"step": 23400
},
{
"epoch": 0.188,
"grad_norm": 4.740218639373779,
"learning_rate": 8.153654618473897e-06,
"loss": 0.2731,
"step": 23500
},
{
"epoch": 0.1888,
"grad_norm": 5.427436828613281,
"learning_rate": 8.145622489959841e-06,
"loss": 0.269,
"step": 23600
},
{
"epoch": 0.1896,
"grad_norm": 13.593464851379395,
"learning_rate": 8.137590361445785e-06,
"loss": 0.263,
"step": 23700
},
{
"epoch": 0.1904,
"grad_norm": 7.22516393661499,
"learning_rate": 8.129558232931729e-06,
"loss": 0.3125,
"step": 23800
},
{
"epoch": 0.1912,
"grad_norm": 14.015786170959473,
"learning_rate": 8.121526104417673e-06,
"loss": 0.2832,
"step": 23900
},
{
"epoch": 0.192,
"grad_norm": 8.892059326171875,
"learning_rate": 8.113493975903615e-06,
"loss": 0.3009,
"step": 24000
},
{
"epoch": 0.192,
"eval_test1_cer": 0.07456214056314886,
"eval_test1_cer_norm": 0.05662352630441569,
"eval_test1_loss": 0.22622939944267273,
"eval_test1_runtime": 3568.0234,
"eval_test1_samples_per_second": 0.701,
"eval_test1_steps_per_second": 0.175,
"eval_test1_wer": 0.19636140995364296,
"eval_test1_wer_norm": 0.13466202986644846,
"step": 24000
},
{
"epoch": 0.192,
"eval_test2_cer": 0.17353100384514877,
"eval_test2_cer_norm": 0.14121571893399443,
"eval_test2_loss": 0.37227049469947815,
"eval_test2_runtime": 3757.9926,
"eval_test2_samples_per_second": 0.665,
"eval_test2_steps_per_second": 0.166,
"eval_test2_wer": 0.3350022888532845,
"eval_test2_wer_norm": 0.2660726564290626,
"step": 24000
},
{
"epoch": 0.1928,
"grad_norm": 6.917489051818848,
"learning_rate": 8.105461847389559e-06,
"loss": 0.2745,
"step": 24100
},
{
"epoch": 0.1936,
"grad_norm": 9.797795295715332,
"learning_rate": 8.097429718875503e-06,
"loss": 0.2792,
"step": 24200
},
{
"epoch": 0.1944,
"grad_norm": 7.5331315994262695,
"learning_rate": 8.089397590361447e-06,
"loss": 0.2631,
"step": 24300
},
{
"epoch": 0.1952,
"grad_norm": 28.057600021362305,
"learning_rate": 8.08136546184739e-06,
"loss": 0.2752,
"step": 24400
},
{
"epoch": 0.196,
"grad_norm": 6.589796543121338,
"learning_rate": 8.073413654618475e-06,
"loss": 0.2601,
"step": 24500
},
{
"epoch": 0.1968,
"grad_norm": 7.585195541381836,
"learning_rate": 8.065381526104419e-06,
"loss": 0.2845,
"step": 24600
},
{
"epoch": 0.1976,
"grad_norm": 12.819061279296875,
"learning_rate": 8.057349397590361e-06,
"loss": 0.2807,
"step": 24700
},
{
"epoch": 0.1984,
"grad_norm": 6.550510883331299,
"learning_rate": 8.049317269076305e-06,
"loss": 0.2641,
"step": 24800
},
{
"epoch": 0.1992,
"grad_norm": 4.5952653884887695,
"learning_rate": 8.041285140562249e-06,
"loss": 0.2773,
"step": 24900
},
{
"epoch": 0.2,
"grad_norm": 7.8463134765625,
"learning_rate": 8.033253012048193e-06,
"loss": 0.299,
"step": 25000
},
{
"epoch": 0.2008,
"grad_norm": 8.69621753692627,
"learning_rate": 8.025220883534137e-06,
"loss": 0.2568,
"step": 25100
},
{
"epoch": 0.2016,
"grad_norm": 19.935726165771484,
"learning_rate": 8.01718875502008e-06,
"loss": 0.2875,
"step": 25200
},
{
"epoch": 0.2024,
"grad_norm": 7.2960662841796875,
"learning_rate": 8.009156626506025e-06,
"loss": 0.2757,
"step": 25300
},
{
"epoch": 0.2032,
"grad_norm": 14.16852855682373,
"learning_rate": 8.001124497991969e-06,
"loss": 0.2733,
"step": 25400
},
{
"epoch": 0.204,
"grad_norm": 10.322851181030273,
"learning_rate": 7.993092369477912e-06,
"loss": 0.2732,
"step": 25500
},
{
"epoch": 0.2048,
"grad_norm": 6.632780075073242,
"learning_rate": 7.985060240963856e-06,
"loss": 0.2705,
"step": 25600
},
{
"epoch": 0.2056,
"grad_norm": 9.688406944274902,
"learning_rate": 7.9770281124498e-06,
"loss": 0.3111,
"step": 25700
},
{
"epoch": 0.2064,
"grad_norm": 11.941987037658691,
"learning_rate": 7.968995983935744e-06,
"loss": 0.2688,
"step": 25800
},
{
"epoch": 0.2072,
"grad_norm": 6.580658912658691,
"learning_rate": 7.960963855421688e-06,
"loss": 0.2842,
"step": 25900
},
{
"epoch": 0.208,
"grad_norm": 9.130199432373047,
"learning_rate": 7.95293172690763e-06,
"loss": 0.2789,
"step": 26000
},
{
"epoch": 0.2088,
"grad_norm": 7.189250469207764,
"learning_rate": 7.944899598393574e-06,
"loss": 0.2687,
"step": 26100
},
{
"epoch": 0.2096,
"grad_norm": 6.851436614990234,
"learning_rate": 7.936867469879518e-06,
"loss": 0.3035,
"step": 26200
},
{
"epoch": 0.2104,
"grad_norm": 5.916522979736328,
"learning_rate": 7.928835341365462e-06,
"loss": 0.2895,
"step": 26300
},
{
"epoch": 0.2112,
"grad_norm": 9.012747764587402,
"learning_rate": 7.920803212851406e-06,
"loss": 0.258,
"step": 26400
},
{
"epoch": 0.212,
"grad_norm": 4.732649803161621,
"learning_rate": 7.91277108433735e-06,
"loss": 0.2707,
"step": 26500
},
{
"epoch": 0.2128,
"grad_norm": 7.508853912353516,
"learning_rate": 7.904738955823294e-06,
"loss": 0.2985,
"step": 26600
},
{
"epoch": 0.2136,
"grad_norm": 7.832762241363525,
"learning_rate": 7.896706827309238e-06,
"loss": 0.2799,
"step": 26700
},
{
"epoch": 0.2144,
"grad_norm": 11.426980972290039,
"learning_rate": 7.888674698795182e-06,
"loss": 0.2877,
"step": 26800
},
{
"epoch": 0.2152,
"grad_norm": 9.61413288116455,
"learning_rate": 7.880642570281126e-06,
"loss": 0.2869,
"step": 26900
},
{
"epoch": 0.216,
"grad_norm": 11.701996803283691,
"learning_rate": 7.87261044176707e-06,
"loss": 0.2607,
"step": 27000
},
{
"epoch": 0.2168,
"grad_norm": 7.202447891235352,
"learning_rate": 7.864578313253013e-06,
"loss": 0.2763,
"step": 27100
},
{
"epoch": 0.2176,
"grad_norm": 9.715481758117676,
"learning_rate": 7.856546184738957e-06,
"loss": 0.2871,
"step": 27200
},
{
"epoch": 0.2184,
"grad_norm": 8.603879928588867,
"learning_rate": 7.848514056224901e-06,
"loss": 0.2876,
"step": 27300
},
{
"epoch": 0.2192,
"grad_norm": 7.729073524475098,
"learning_rate": 7.840481927710844e-06,
"loss": 0.25,
"step": 27400
},
{
"epoch": 0.22,
"grad_norm": 4.843418598175049,
"learning_rate": 7.832449799196787e-06,
"loss": 0.2913,
"step": 27500
},
{
"epoch": 0.2208,
"grad_norm": 13.062045097351074,
"learning_rate": 7.824417670682731e-06,
"loss": 0.2676,
"step": 27600
},
{
"epoch": 0.2216,
"grad_norm": 9.212933540344238,
"learning_rate": 7.816385542168675e-06,
"loss": 0.2814,
"step": 27700
},
{
"epoch": 0.2224,
"grad_norm": 8.371397972106934,
"learning_rate": 7.80835341365462e-06,
"loss": 0.2556,
"step": 27800
},
{
"epoch": 0.2232,
"grad_norm": 11.034435272216797,
"learning_rate": 7.800321285140563e-06,
"loss": 0.2656,
"step": 27900
},
{
"epoch": 0.224,
"grad_norm": 7.741024494171143,
"learning_rate": 7.792289156626507e-06,
"loss": 0.2481,
"step": 28000
},
{
"epoch": 0.224,
"eval_test1_cer": 0.08188158189558593,
"eval_test1_cer_norm": 0.06053257137368838,
"eval_test1_loss": 0.21281211078166962,
"eval_test1_runtime": 3606.8392,
"eval_test1_samples_per_second": 0.693,
"eval_test1_steps_per_second": 0.173,
"eval_test1_wer": 0.20149275489081314,
"eval_test1_wer_norm": 0.13837341827639615,
"step": 28000
},
{
"epoch": 0.224,
"eval_test2_cer": 0.22203867547691045,
"eval_test2_cer_norm": 0.17609234583204214,
"eval_test2_loss": 0.35916781425476074,
"eval_test2_runtime": 3907.2621,
"eval_test2_samples_per_second": 0.64,
"eval_test2_steps_per_second": 0.16,
"eval_test2_wer": 0.3924525062943465,
"eval_test2_wer_norm": 0.32526358010543205,
"step": 28000
},
{
"epoch": 0.2248,
"grad_norm": 8.185860633850098,
"learning_rate": 7.784257028112451e-06,
"loss": 0.2649,
"step": 28100
},
{
"epoch": 0.2256,
"grad_norm": 9.275001525878906,
"learning_rate": 7.776224899598395e-06,
"loss": 0.2835,
"step": 28200
},
{
"epoch": 0.2264,
"grad_norm": 11.974363327026367,
"learning_rate": 7.768192771084337e-06,
"loss": 0.279,
"step": 28300
},
{
"epoch": 0.2272,
"grad_norm": 8.101808547973633,
"learning_rate": 7.760160642570281e-06,
"loss": 0.2934,
"step": 28400
},
{
"epoch": 0.228,
"grad_norm": 6.571002006530762,
"learning_rate": 7.752128514056225e-06,
"loss": 0.2728,
"step": 28500
},
{
"epoch": 0.2288,
"grad_norm": 5.396939277648926,
"learning_rate": 7.744096385542169e-06,
"loss": 0.2606,
"step": 28600
},
{
"epoch": 0.2296,
"grad_norm": 11.369711875915527,
"learning_rate": 7.736144578313253e-06,
"loss": 0.2621,
"step": 28700
},
{
"epoch": 0.2304,
"grad_norm": 11.154069900512695,
"learning_rate": 7.728112449799197e-06,
"loss": 0.2668,
"step": 28800
},
{
"epoch": 0.2312,
"grad_norm": 9.739912033081055,
"learning_rate": 7.720080321285141e-06,
"loss": 0.2448,
"step": 28900
},
{
"epoch": 0.232,
"grad_norm": 7.0749125480651855,
"learning_rate": 7.712048192771085e-06,
"loss": 0.2571,
"step": 29000
},
{
"epoch": 0.2328,
"grad_norm": 11.714834213256836,
"learning_rate": 7.704016064257029e-06,
"loss": 0.2687,
"step": 29100
},
{
"epoch": 0.2336,
"grad_norm": 7.279887676239014,
"learning_rate": 7.695983935742973e-06,
"loss": 0.2606,
"step": 29200
},
{
"epoch": 0.2344,
"grad_norm": 13.882575035095215,
"learning_rate": 7.687951807228917e-06,
"loss": 0.2906,
"step": 29300
},
{
"epoch": 0.2352,
"grad_norm": 11.140130996704102,
"learning_rate": 7.67991967871486e-06,
"loss": 0.2795,
"step": 29400
},
{
"epoch": 0.236,
"grad_norm": 8.242719650268555,
"learning_rate": 7.671887550200805e-06,
"loss": 0.2531,
"step": 29500
},
{
"epoch": 0.2368,
"grad_norm": 3.9837567806243896,
"learning_rate": 7.663855421686748e-06,
"loss": 0.2684,
"step": 29600
},
{
"epoch": 0.2376,
"grad_norm": 4.7640275955200195,
"learning_rate": 7.655823293172692e-06,
"loss": 0.2812,
"step": 29700
},
{
"epoch": 0.2384,
"grad_norm": 5.323102951049805,
"learning_rate": 7.647791164658636e-06,
"loss": 0.2742,
"step": 29800
},
{
"epoch": 0.2392,
"grad_norm": 5.6695966720581055,
"learning_rate": 7.63975903614458e-06,
"loss": 0.2662,
"step": 29900
},
{
"epoch": 0.24,
"grad_norm": 11.021814346313477,
"learning_rate": 7.631726907630522e-06,
"loss": 0.2591,
"step": 30000
},
{
"epoch": 0.2408,
"grad_norm": 7.066900253295898,
"learning_rate": 7.623694779116466e-06,
"loss": 0.2799,
"step": 30100
},
{
"epoch": 0.2416,
"grad_norm": 7.006994247436523,
"learning_rate": 7.61566265060241e-06,
"loss": 0.2763,
"step": 30200
},
{
"epoch": 0.2424,
"grad_norm": 6.179232597351074,
"learning_rate": 7.607630522088354e-06,
"loss": 0.2607,
"step": 30300
},
{
"epoch": 0.2432,
"grad_norm": 11.01894760131836,
"learning_rate": 7.599598393574298e-06,
"loss": 0.2892,
"step": 30400
},
{
"epoch": 0.244,
"grad_norm": 8.360103607177734,
"learning_rate": 7.591566265060242e-06,
"loss": 0.2682,
"step": 30500
},
{
"epoch": 0.2448,
"grad_norm": 12.027982711791992,
"learning_rate": 7.583614457831326e-06,
"loss": 0.2795,
"step": 30600
},
{
"epoch": 0.2456,
"grad_norm": 9.705676078796387,
"learning_rate": 7.5755823293172694e-06,
"loss": 0.2604,
"step": 30700
},
{
"epoch": 0.2464,
"grad_norm": 78.23721313476562,
"learning_rate": 7.567550200803213e-06,
"loss": 0.2744,
"step": 30800
},
{
"epoch": 0.2472,
"grad_norm": 7.072327613830566,
"learning_rate": 7.559518072289157e-06,
"loss": 0.2518,
"step": 30900
},
{
"epoch": 0.248,
"grad_norm": 6.551907062530518,
"learning_rate": 7.551485943775101e-06,
"loss": 0.252,
"step": 31000
},
{
"epoch": 0.2488,
"grad_norm": 7.861015796661377,
"learning_rate": 7.543453815261045e-06,
"loss": 0.2616,
"step": 31100
},
{
"epoch": 0.2496,
"grad_norm": 13.113914489746094,
"learning_rate": 7.535421686746989e-06,
"loss": 0.2929,
"step": 31200
},
{
"epoch": 0.2504,
"grad_norm": 10.613142967224121,
"learning_rate": 7.527389558232933e-06,
"loss": 0.2603,
"step": 31300
},
{
"epoch": 0.2512,
"grad_norm": 9.165702819824219,
"learning_rate": 7.519357429718877e-06,
"loss": 0.2633,
"step": 31400
},
{
"epoch": 0.252,
"grad_norm": 5.82572078704834,
"learning_rate": 7.511325301204821e-06,
"loss": 0.2553,
"step": 31500
},
{
"epoch": 0.2528,
"grad_norm": 4.581362247467041,
"learning_rate": 7.503293172690763e-06,
"loss": 0.2735,
"step": 31600
},
{
"epoch": 0.2536,
"grad_norm": 6.59524393081665,
"learning_rate": 7.495261044176707e-06,
"loss": 0.2667,
"step": 31700
},
{
"epoch": 0.2544,
"grad_norm": 10.484724044799805,
"learning_rate": 7.487228915662651e-06,
"loss": 0.2679,
"step": 31800
},
{
"epoch": 0.2552,
"grad_norm": 8.551301002502441,
"learning_rate": 7.479196787148595e-06,
"loss": 0.2538,
"step": 31900
},
{
"epoch": 0.256,
"grad_norm": 6.701406002044678,
"learning_rate": 7.471164658634539e-06,
"loss": 0.2507,
"step": 32000
},
{
"epoch": 0.256,
"eval_test1_cer": 0.08450033609679587,
"eval_test1_cer_norm": 0.058093019905395345,
"eval_test1_loss": 0.20838095247745514,
"eval_test1_runtime": 3620.5796,
"eval_test1_samples_per_second": 0.69,
"eval_test1_steps_per_second": 0.173,
"eval_test1_wer": 0.20155106562873554,
"eval_test1_wer_norm": 0.13854875946111808,
"step": 32000
},
{
"epoch": 0.256,
"eval_test2_cer": 0.17676018217792214,
"eval_test2_cer_norm": 0.13646575766966224,
"eval_test2_loss": 0.35614562034606934,
"eval_test2_runtime": 3787.2574,
"eval_test2_samples_per_second": 0.66,
"eval_test2_steps_per_second": 0.165,
"eval_test2_wer": 0.3315690089265278,
"eval_test2_wer_norm": 0.26286385514554206,
"step": 32000
},
{
"epoch": 0.2568,
"grad_norm": 9.485516548156738,
"learning_rate": 7.463132530120483e-06,
"loss": 0.2773,
"step": 32100
},
{
"epoch": 0.2576,
"grad_norm": 7.148492336273193,
"learning_rate": 7.4551004016064265e-06,
"loss": 0.2609,
"step": 32200
},
{
"epoch": 0.2584,
"grad_norm": 7.948647499084473,
"learning_rate": 7.4470682730923705e-06,
"loss": 0.2901,
"step": 32300
},
{
"epoch": 0.2592,
"grad_norm": 11.446524620056152,
"learning_rate": 7.439036144578314e-06,
"loss": 0.2531,
"step": 32400
},
{
"epoch": 0.26,
"grad_norm": 7.954312324523926,
"learning_rate": 7.4310040160642574e-06,
"loss": 0.2474,
"step": 32500
},
{
"epoch": 0.2608,
"grad_norm": 6.527101039886475,
"learning_rate": 7.422971887550201e-06,
"loss": 0.2713,
"step": 32600
},
{
"epoch": 0.2616,
"grad_norm": 4.255068302154541,
"learning_rate": 7.414939759036145e-06,
"loss": 0.2774,
"step": 32700
},
{
"epoch": 0.2624,
"grad_norm": 12.29005241394043,
"learning_rate": 7.406907630522089e-06,
"loss": 0.2869,
"step": 32800
},
{
"epoch": 0.2632,
"grad_norm": 7.255998611450195,
"learning_rate": 7.398875502008033e-06,
"loss": 0.2776,
"step": 32900
},
{
"epoch": 0.264,
"grad_norm": 5.739976406097412,
"learning_rate": 7.390843373493977e-06,
"loss": 0.2658,
"step": 33000
},
{
"epoch": 0.2648,
"grad_norm": 9.824767112731934,
"learning_rate": 7.38281124497992e-06,
"loss": 0.2741,
"step": 33100
},
{
"epoch": 0.2656,
"grad_norm": 5.075085163116455,
"learning_rate": 7.374779116465864e-06,
"loss": 0.2626,
"step": 33200
},
{
"epoch": 0.2664,
"grad_norm": 11.58659839630127,
"learning_rate": 7.366746987951808e-06,
"loss": 0.2834,
"step": 33300
},
{
"epoch": 0.2672,
"grad_norm": 3.7526087760925293,
"learning_rate": 7.358795180722892e-06,
"loss": 0.2896,
"step": 33400
},
{
"epoch": 0.268,
"grad_norm": 5.333492755889893,
"learning_rate": 7.350763052208836e-06,
"loss": 0.2575,
"step": 33500
},
{
"epoch": 0.2688,
"grad_norm": 8.13162612915039,
"learning_rate": 7.34273092369478e-06,
"loss": 0.2513,
"step": 33600
},
{
"epoch": 0.2696,
"grad_norm": 8.09466552734375,
"learning_rate": 7.334698795180723e-06,
"loss": 0.3005,
"step": 33700
},
{
"epoch": 0.2704,
"grad_norm": 4.5985331535339355,
"learning_rate": 7.326666666666667e-06,
"loss": 0.3217,
"step": 33800
},
{
"epoch": 0.2712,
"grad_norm": 6.758331775665283,
"learning_rate": 7.318634538152611e-06,
"loss": 0.3054,
"step": 33900
},
{
"epoch": 0.272,
"grad_norm": 9.720072746276855,
"learning_rate": 7.310602409638555e-06,
"loss": 0.2554,
"step": 34000
},
{
"epoch": 0.2728,
"grad_norm": 10.393074989318848,
"learning_rate": 7.302570281124498e-06,
"loss": 0.2348,
"step": 34100
},
{
"epoch": 0.2736,
"grad_norm": 5.730389595031738,
"learning_rate": 7.294538152610442e-06,
"loss": 0.264,
"step": 34200
},
{
"epoch": 0.2744,
"grad_norm": 9.347960472106934,
"learning_rate": 7.286506024096386e-06,
"loss": 0.3121,
"step": 34300
},
{
"epoch": 0.2752,
"grad_norm": 5.9382171630859375,
"learning_rate": 7.27847389558233e-06,
"loss": 0.256,
"step": 34400
},
{
"epoch": 0.276,
"grad_norm": 13.730400085449219,
"learning_rate": 7.270441767068274e-06,
"loss": 0.2474,
"step": 34500
},
{
"epoch": 0.2768,
"grad_norm": 5.6664347648620605,
"learning_rate": 7.262409638554218e-06,
"loss": 0.2642,
"step": 34600
},
{
"epoch": 0.2776,
"grad_norm": 4.812320232391357,
"learning_rate": 7.2543775100401615e-06,
"loss": 0.2697,
"step": 34700
},
{
"epoch": 0.2784,
"grad_norm": 6.524253845214844,
"learning_rate": 7.2463453815261055e-06,
"loss": 0.2511,
"step": 34800
},
{
"epoch": 0.2792,
"grad_norm": 10.69779109954834,
"learning_rate": 7.238313253012049e-06,
"loss": 0.2455,
"step": 34900
},
{
"epoch": 0.28,
"grad_norm": 6.162754535675049,
"learning_rate": 7.230281124497992e-06,
"loss": 0.3028,
"step": 35000
},
{
"epoch": 0.2808,
"grad_norm": 10.023324966430664,
"learning_rate": 7.2222489959839355e-06,
"loss": 0.2618,
"step": 35100
},
{
"epoch": 0.2816,
"grad_norm": 7.126629829406738,
"learning_rate": 7.2142168674698794e-06,
"loss": 0.2482,
"step": 35200
},
{
"epoch": 0.2824,
"grad_norm": 5.309641361236572,
"learning_rate": 7.206184738955823e-06,
"loss": 0.2641,
"step": 35300
},
{
"epoch": 0.2832,
"grad_norm": 8.043412208557129,
"learning_rate": 7.198152610441767e-06,
"loss": 0.2779,
"step": 35400
},
{
"epoch": 0.284,
"grad_norm": 6.982008934020996,
"learning_rate": 7.190120481927711e-06,
"loss": 0.2519,
"step": 35500
},
{
"epoch": 0.2848,
"grad_norm": 14.100213050842285,
"learning_rate": 7.182088353413655e-06,
"loss": 0.2621,
"step": 35600
},
{
"epoch": 0.2856,
"grad_norm": 7.013679027557373,
"learning_rate": 7.174056224899599e-06,
"loss": 0.2511,
"step": 35700
},
{
"epoch": 0.2864,
"grad_norm": 12.395895004272461,
"learning_rate": 7.166024096385543e-06,
"loss": 0.2574,
"step": 35800
},
{
"epoch": 0.2872,
"grad_norm": 5.958752155303955,
"learning_rate": 7.157991967871486e-06,
"loss": 0.2376,
"step": 35900
},
{
"epoch": 0.288,
"grad_norm": 7.98759126663208,
"learning_rate": 7.14995983935743e-06,
"loss": 0.2558,
"step": 36000
},
{
"epoch": 0.288,
"eval_test1_cer": 0.09376167002763462,
"eval_test1_cer_norm": 0.06997382764664922,
"eval_test1_loss": 0.2042306363582611,
"eval_test1_runtime": 3524.6047,
"eval_test1_samples_per_second": 0.709,
"eval_test1_steps_per_second": 0.177,
"eval_test1_wer": 0.2142628064958162,
"eval_test1_wer_norm": 0.15251760717729917,
"step": 36000
},
{
"epoch": 0.288,
"eval_test2_cer": 0.18322787172882368,
"eval_test2_cer_norm": 0.1433607065385807,
"eval_test2_loss": 0.34622320532798767,
"eval_test2_runtime": 2531.2967,
"eval_test2_samples_per_second": 0.988,
"eval_test2_steps_per_second": 0.247,
"eval_test2_wer": 0.3355745021744106,
"eval_test2_wer_norm": 0.2676197570479028,
"step": 36000
},
{
"epoch": 0.2888,
"grad_norm": 7.333773612976074,
"learning_rate": 7.141927710843374e-06,
"loss": 0.2581,
"step": 36100
},
{
"epoch": 0.2896,
"grad_norm": 6.082370758056641,
"learning_rate": 7.133895582329318e-06,
"loss": 0.2444,
"step": 36200
},
{
"epoch": 0.2904,
"grad_norm": 7.414235591888428,
"learning_rate": 7.125863453815262e-06,
"loss": 0.2501,
"step": 36300
},
{
"epoch": 0.2912,
"grad_norm": 7.523865699768066,
"learning_rate": 7.117831325301206e-06,
"loss": 0.2732,
"step": 36400
},
{
"epoch": 0.292,
"grad_norm": 4.851339817047119,
"learning_rate": 7.1097991967871496e-06,
"loss": 0.2674,
"step": 36500
},
{
"epoch": 0.2928,
"grad_norm": 14.28159236907959,
"learning_rate": 7.1017670682730935e-06,
"loss": 0.2652,
"step": 36600
},
{
"epoch": 0.2936,
"grad_norm": 9.865965843200684,
"learning_rate": 7.093734939759037e-06,
"loss": 0.2586,
"step": 36700
},
{
"epoch": 0.2944,
"grad_norm": 8.199728965759277,
"learning_rate": 7.085702811244981e-06,
"loss": 0.2606,
"step": 36800
},
{
"epoch": 0.2952,
"grad_norm": 6.126343727111816,
"learning_rate": 7.0776706827309235e-06,
"loss": 0.2244,
"step": 36900
},
{
"epoch": 0.296,
"grad_norm": 12.142101287841797,
"learning_rate": 7.0696385542168675e-06,
"loss": 0.2737,
"step": 37000
},
{
"epoch": 0.2968,
"grad_norm": 18.582122802734375,
"learning_rate": 7.061606425702811e-06,
"loss": 0.2617,
"step": 37100
},
{
"epoch": 0.2976,
"grad_norm": 6.676758766174316,
"learning_rate": 7.053574297188755e-06,
"loss": 0.2679,
"step": 37200
},
{
"epoch": 0.2984,
"grad_norm": 11.17313003540039,
"learning_rate": 7.045542168674699e-06,
"loss": 0.2896,
"step": 37300
},
{
"epoch": 0.2992,
"grad_norm": 6.7832159996032715,
"learning_rate": 7.037510040160643e-06,
"loss": 0.2645,
"step": 37400
},
{
"epoch": 0.3,
"grad_norm": 7.81158971786499,
"learning_rate": 7.029558232931728e-06,
"loss": 0.2486,
"step": 37500
},
{
"epoch": 0.3008,
"grad_norm": 5.17408561706543,
"learning_rate": 7.0215261044176705e-06,
"loss": 0.274,
"step": 37600
},
{
"epoch": 0.3016,
"grad_norm": 4.426403045654297,
"learning_rate": 7.0134939759036144e-06,
"loss": 0.2574,
"step": 37700
},
{
"epoch": 0.3024,
"grad_norm": 4.935957908630371,
"learning_rate": 7.005461847389558e-06,
"loss": 0.239,
"step": 37800
},
{
"epoch": 0.3032,
"grad_norm": 5.82489013671875,
"learning_rate": 6.997429718875502e-06,
"loss": 0.2549,
"step": 37900
},
{
"epoch": 0.304,
"grad_norm": 5.054333209991455,
"learning_rate": 6.989397590361446e-06,
"loss": 0.2675,
"step": 38000
},
{
"epoch": 0.3048,
"grad_norm": 9.367691040039062,
"learning_rate": 6.98136546184739e-06,
"loss": 0.2699,
"step": 38100
},
{
"epoch": 0.3056,
"grad_norm": 11.023392677307129,
"learning_rate": 6.973333333333334e-06,
"loss": 0.2483,
"step": 38200
},
{
"epoch": 0.3064,
"grad_norm": 12.55837345123291,
"learning_rate": 6.965301204819278e-06,
"loss": 0.2557,
"step": 38300
},
{
"epoch": 0.3072,
"grad_norm": 5.479390621185303,
"learning_rate": 6.957269076305222e-06,
"loss": 0.2625,
"step": 38400
},
{
"epoch": 0.308,
"grad_norm": 11.671854972839355,
"learning_rate": 6.949236947791165e-06,
"loss": 0.2608,
"step": 38500
},
{
"epoch": 0.3088,
"grad_norm": 5.811091423034668,
"learning_rate": 6.941204819277109e-06,
"loss": 0.2671,
"step": 38600
},
{
"epoch": 0.3096,
"grad_norm": 5.157212734222412,
"learning_rate": 6.933172690763053e-06,
"loss": 0.2701,
"step": 38700
},
{
"epoch": 0.3104,
"grad_norm": 6.742706775665283,
"learning_rate": 6.925140562248997e-06,
"loss": 0.264,
"step": 38800
},
{
"epoch": 0.3112,
"grad_norm": 10.843968391418457,
"learning_rate": 6.917108433734941e-06,
"loss": 0.2476,
"step": 38900
},
{
"epoch": 0.312,
"grad_norm": 5.495096206665039,
"learning_rate": 6.9090763052208846e-06,
"loss": 0.2674,
"step": 39000
},
{
"epoch": 0.3128,
"grad_norm": 11.919147491455078,
"learning_rate": 6.901124497991969e-06,
"loss": 0.2397,
"step": 39100
},
{
"epoch": 0.3136,
"grad_norm": 9.006637573242188,
"learning_rate": 6.893092369477912e-06,
"loss": 0.2502,
"step": 39200
},
{
"epoch": 0.3144,
"grad_norm": 4.003146171569824,
"learning_rate": 6.885060240963856e-06,
"loss": 0.2697,
"step": 39300
},
{
"epoch": 0.3152,
"grad_norm": 10.066797256469727,
"learning_rate": 6.8770281124498e-06,
"loss": 0.2295,
"step": 39400
},
{
"epoch": 0.316,
"grad_norm": 8.48736572265625,
"learning_rate": 6.868995983935744e-06,
"loss": 0.2566,
"step": 39500
},
{
"epoch": 0.3168,
"grad_norm": 6.740108966827393,
"learning_rate": 6.860963855421688e-06,
"loss": 0.2529,
"step": 39600
},
{
"epoch": 0.3176,
"grad_norm": 7.446850299835205,
"learning_rate": 6.8529317269076315e-06,
"loss": 0.2571,
"step": 39700
},
{
"epoch": 0.3184,
"grad_norm": 4.642506122589111,
"learning_rate": 6.8448995983935755e-06,
"loss": 0.3017,
"step": 39800
},
{
"epoch": 0.3192,
"grad_norm": 5.234320163726807,
"learning_rate": 6.836867469879519e-06,
"loss": 0.2668,
"step": 39900
},
{
"epoch": 0.32,
"grad_norm": 7.142499923706055,
"learning_rate": 6.828835341365463e-06,
"loss": 0.2566,
"step": 40000
},
{
"epoch": 0.32,
"eval_test1_cer": 0.06914724774068265,
"eval_test1_cer_norm": 0.04478113669652076,
"eval_test1_loss": 0.19836583733558655,
"eval_test1_runtime": 2432.2319,
"eval_test1_samples_per_second": 1.028,
"eval_test1_steps_per_second": 0.257,
"eval_test1_wer": 0.17665238053587567,
"eval_test1_wer_norm": 0.11657266430930185,
"step": 40000
},
{
"epoch": 0.32,
"eval_test2_cer": 0.15002146563631613,
"eval_test2_cer_norm": 0.10912806011775643,
"eval_test2_loss": 0.33799564838409424,
"eval_test2_runtime": 2481.3184,
"eval_test2_samples_per_second": 1.008,
"eval_test2_steps_per_second": 0.252,
"eval_test2_wer": 0.28587777523460745,
"eval_test2_wer_norm": 0.2201466880586752,
"step": 40000
},
{
"epoch": 0.3208,
"grad_norm": 6.141145706176758,
"learning_rate": 6.8208032128514055e-06,
"loss": 0.2562,
"step": 40100
},
{
"epoch": 0.3216,
"grad_norm": 5.139159679412842,
"learning_rate": 6.8127710843373495e-06,
"loss": 0.2703,
"step": 40200
},
{
"epoch": 0.3224,
"grad_norm": 7.440242767333984,
"learning_rate": 6.804738955823293e-06,
"loss": 0.2821,
"step": 40300
},
{
"epoch": 0.3232,
"grad_norm": 6.795882225036621,
"learning_rate": 6.796706827309237e-06,
"loss": 0.2452,
"step": 40400
},
{
"epoch": 0.324,
"grad_norm": 7.417604446411133,
"learning_rate": 6.788674698795181e-06,
"loss": 0.2521,
"step": 40500
},
{
"epoch": 0.3248,
"grad_norm": 6.449469089508057,
"learning_rate": 6.780642570281125e-06,
"loss": 0.2813,
"step": 40600
},
{
"epoch": 0.3256,
"grad_norm": 10.74730110168457,
"learning_rate": 6.772610441767069e-06,
"loss": 0.2551,
"step": 40700
},
{
"epoch": 0.3264,
"grad_norm": 3.9926106929779053,
"learning_rate": 6.764578313253013e-06,
"loss": 0.242,
"step": 40800
},
{
"epoch": 0.3272,
"grad_norm": 3.6253511905670166,
"learning_rate": 6.756546184738957e-06,
"loss": 0.2359,
"step": 40900
},
{
"epoch": 0.328,
"grad_norm": 7.051224231719971,
"learning_rate": 6.7485140562249e-06,
"loss": 0.2618,
"step": 41000
},
{
"epoch": 0.3288,
"grad_norm": 9.776261329650879,
"learning_rate": 6.740481927710844e-06,
"loss": 0.2649,
"step": 41100
},
{
"epoch": 0.3296,
"grad_norm": 5.289026260375977,
"learning_rate": 6.732449799196788e-06,
"loss": 0.2446,
"step": 41200
},
{
"epoch": 0.3304,
"grad_norm": 9.757101058959961,
"learning_rate": 6.724417670682732e-06,
"loss": 0.2605,
"step": 41300
},
{
"epoch": 0.3312,
"grad_norm": 7.856915473937988,
"learning_rate": 6.716385542168675e-06,
"loss": 0.2566,
"step": 41400
},
{
"epoch": 0.332,
"grad_norm": 4.344404697418213,
"learning_rate": 6.708353413654619e-06,
"loss": 0.2482,
"step": 41500
},
{
"epoch": 0.3328,
"grad_norm": 7.893800735473633,
"learning_rate": 6.700321285140563e-06,
"loss": 0.2555,
"step": 41600
},
{
"epoch": 0.3336,
"grad_norm": 10.901208877563477,
"learning_rate": 6.6922891566265066e-06,
"loss": 0.2455,
"step": 41700
},
{
"epoch": 0.3344,
"grad_norm": 8.60644817352295,
"learning_rate": 6.6842570281124505e-06,
"loss": 0.242,
"step": 41800
},
{
"epoch": 0.3352,
"grad_norm": 10.857059478759766,
"learning_rate": 6.6762248995983936e-06,
"loss": 0.2611,
"step": 41900
},
{
"epoch": 0.336,
"grad_norm": 4.682589054107666,
"learning_rate": 6.6681927710843375e-06,
"loss": 0.2365,
"step": 42000
},
{
"epoch": 0.3368,
"grad_norm": 10.836357116699219,
"learning_rate": 6.660160642570281e-06,
"loss": 0.2477,
"step": 42100
},
{
"epoch": 0.3376,
"grad_norm": 4.692202568054199,
"learning_rate": 6.652128514056225e-06,
"loss": 0.2494,
"step": 42200
},
{
"epoch": 0.3384,
"grad_norm": 10.428234100341797,
"learning_rate": 6.644096385542169e-06,
"loss": 0.2605,
"step": 42300
},
{
"epoch": 0.3392,
"grad_norm": 4.078035831451416,
"learning_rate": 6.636064257028113e-06,
"loss": 0.2358,
"step": 42400
},
{
"epoch": 0.34,
"grad_norm": 7.381161689758301,
"learning_rate": 6.628032128514057e-06,
"loss": 0.2371,
"step": 42500
},
{
"epoch": 0.3408,
"grad_norm": 2.6584036350250244,
"learning_rate": 6.620000000000001e-06,
"loss": 0.2795,
"step": 42600
},
{
"epoch": 0.3416,
"grad_norm": 10.325010299682617,
"learning_rate": 6.611967871485945e-06,
"loss": 0.2476,
"step": 42700
},
{
"epoch": 0.3424,
"grad_norm": 6.986746788024902,
"learning_rate": 6.603935742971887e-06,
"loss": 0.2354,
"step": 42800
},
{
"epoch": 0.3432,
"grad_norm": 6.040923118591309,
"learning_rate": 6.595903614457831e-06,
"loss": 0.2748,
"step": 42900
},
{
"epoch": 0.344,
"grad_norm": 4.043920993804932,
"learning_rate": 6.587871485943775e-06,
"loss": 0.2579,
"step": 43000
},
{
"epoch": 0.3448,
"grad_norm": 7.066967964172363,
"learning_rate": 6.57991967871486e-06,
"loss": 0.2558,
"step": 43100
},
{
"epoch": 0.3456,
"grad_norm": 7.57224702835083,
"learning_rate": 6.571887550200804e-06,
"loss": 0.2542,
"step": 43200
},
{
"epoch": 0.3464,
"grad_norm": 11.219292640686035,
"learning_rate": 6.563855421686748e-06,
"loss": 0.2688,
"step": 43300
},
{
"epoch": 0.3472,
"grad_norm": 9.798463821411133,
"learning_rate": 6.555823293172692e-06,
"loss": 0.2555,
"step": 43400
},
{
"epoch": 0.348,
"grad_norm": 6.011776924133301,
"learning_rate": 6.547791164658636e-06,
"loss": 0.2517,
"step": 43500
},
{
"epoch": 0.3488,
"grad_norm": 10.116564750671387,
"learning_rate": 6.539759036144578e-06,
"loss": 0.2585,
"step": 43600
},
{
"epoch": 0.3496,
"grad_norm": 8.014334678649902,
"learning_rate": 6.531807228915663e-06,
"loss": 0.2386,
"step": 43700
},
{
"epoch": 0.3504,
"grad_norm": 6.700802803039551,
"learning_rate": 6.523775100401607e-06,
"loss": 0.2412,
"step": 43800
},
{
"epoch": 0.3512,
"grad_norm": 5.972689151763916,
"learning_rate": 6.515742971887551e-06,
"loss": 0.2461,
"step": 43900
},
{
"epoch": 0.352,
"grad_norm": 6.2880167961120605,
"learning_rate": 6.507710843373495e-06,
"loss": 0.2535,
"step": 44000
},
{
"epoch": 0.352,
"eval_test1_cer": 0.07431940398834865,
"eval_test1_cer_norm": 0.051710807501140536,
"eval_test1_loss": 0.19776684045791626,
"eval_test1_runtime": 2427.0653,
"eval_test1_samples_per_second": 1.03,
"eval_test1_steps_per_second": 0.258,
"eval_test1_wer": 0.17638998221522492,
"eval_test1_wer_norm": 0.11625120547064496,
"step": 44000
},
{
"epoch": 0.352,
"eval_test2_cer": 0.1173377011236794,
"eval_test2_cer_norm": 0.09236519987604587,
"eval_test2_loss": 0.3366641104221344,
"eval_test2_runtime": 2441.4105,
"eval_test2_samples_per_second": 1.024,
"eval_test2_steps_per_second": 0.256,
"eval_test2_wer": 0.2464522774090181,
"eval_test2_wer_norm": 0.1820135228054091,
"step": 44000
},
{
"epoch": 0.3528,
"grad_norm": 13.718138694763184,
"learning_rate": 6.499678714859439e-06,
"loss": 0.2509,
"step": 44100
},
{
"epoch": 0.3536,
"grad_norm": 8.281696319580078,
"learning_rate": 6.491646586345383e-06,
"loss": 0.2489,
"step": 44200
},
{
"epoch": 0.3544,
"grad_norm": 4.6960225105285645,
"learning_rate": 6.483614457831325e-06,
"loss": 0.2693,
"step": 44300
},
{
"epoch": 0.3552,
"grad_norm": 6.965161323547363,
"learning_rate": 6.475582329317269e-06,
"loss": 0.2242,
"step": 44400
},
{
"epoch": 0.356,
"grad_norm": 4.1582722663879395,
"learning_rate": 6.467550200803213e-06,
"loss": 0.2551,
"step": 44500
},
{
"epoch": 0.3568,
"grad_norm": 6.540956020355225,
"learning_rate": 6.459518072289157e-06,
"loss": 0.2504,
"step": 44600
},
{
"epoch": 0.3576,
"grad_norm": 8.521894454956055,
"learning_rate": 6.451485943775101e-06,
"loss": 0.2578,
"step": 44700
},
{
"epoch": 0.3584,
"grad_norm": 9.609855651855469,
"learning_rate": 6.443453815261045e-06,
"loss": 0.2727,
"step": 44800
},
{
"epoch": 0.3592,
"grad_norm": 7.131938457489014,
"learning_rate": 6.4354216867469885e-06,
"loss": 0.2519,
"step": 44900
},
{
"epoch": 0.36,
"grad_norm": 3.9930307865142822,
"learning_rate": 6.4273895582329325e-06,
"loss": 0.246,
"step": 45000
},
{
"epoch": 0.3608,
"grad_norm": 12.197285652160645,
"learning_rate": 6.419357429718876e-06,
"loss": 0.2592,
"step": 45100
},
{
"epoch": 0.3616,
"grad_norm": 11.38996410369873,
"learning_rate": 6.4113253012048195e-06,
"loss": 0.2645,
"step": 45200
},
{
"epoch": 0.3624,
"grad_norm": 9.129228591918945,
"learning_rate": 6.403293172690763e-06,
"loss": 0.254,
"step": 45300
},
{
"epoch": 0.3632,
"grad_norm": 4.596703052520752,
"learning_rate": 6.395261044176707e-06,
"loss": 0.2555,
"step": 45400
},
{
"epoch": 0.364,
"grad_norm": 4.702436923980713,
"learning_rate": 6.387228915662651e-06,
"loss": 0.2617,
"step": 45500
},
{
"epoch": 0.3648,
"grad_norm": 6.283945083618164,
"learning_rate": 6.379196787148595e-06,
"loss": 0.2546,
"step": 45600
},
{
"epoch": 0.3656,
"grad_norm": 8.981857299804688,
"learning_rate": 6.371164658634539e-06,
"loss": 0.2353,
"step": 45700
},
{
"epoch": 0.3664,
"grad_norm": 7.928671360015869,
"learning_rate": 6.363132530120483e-06,
"loss": 0.2657,
"step": 45800
},
{
"epoch": 0.3672,
"grad_norm": 7.3669114112854,
"learning_rate": 6.355100401606427e-06,
"loss": 0.2486,
"step": 45900
},
{
"epoch": 0.368,
"grad_norm": 6.917601108551025,
"learning_rate": 6.347068273092371e-06,
"loss": 0.2524,
"step": 46000
},
{
"epoch": 0.3688,
"grad_norm": 4.770781517028809,
"learning_rate": 6.339036144578313e-06,
"loss": 0.2434,
"step": 46100
},
{
"epoch": 0.3696,
"grad_norm": 12.580122947692871,
"learning_rate": 6.331004016064257e-06,
"loss": 0.2463,
"step": 46200
},
{
"epoch": 0.3704,
"grad_norm": 6.896516799926758,
"learning_rate": 6.322971887550201e-06,
"loss": 0.2365,
"step": 46300
},
{
"epoch": 0.3712,
"grad_norm": 7.098085880279541,
"learning_rate": 6.314939759036145e-06,
"loss": 0.2412,
"step": 46400
},
{
"epoch": 0.372,
"grad_norm": 5.216635704040527,
"learning_rate": 6.306907630522089e-06,
"loss": 0.2209,
"step": 46500
},
{
"epoch": 0.3728,
"grad_norm": 7.694733619689941,
"learning_rate": 6.298875502008033e-06,
"loss": 0.2256,
"step": 46600
},
{
"epoch": 0.3736,
"grad_norm": 37.76133346557617,
"learning_rate": 6.290923694779118e-06,
"loss": 0.24,
"step": 46700
},
{
"epoch": 0.3744,
"grad_norm": 4.706324100494385,
"learning_rate": 6.28289156626506e-06,
"loss": 0.2224,
"step": 46800
},
{
"epoch": 0.3752,
"grad_norm": 9.157281875610352,
"learning_rate": 6.274859437751004e-06,
"loss": 0.2609,
"step": 46900
},
{
"epoch": 0.376,
"grad_norm": 6.885075092315674,
"learning_rate": 6.266827309236948e-06,
"loss": 0.2711,
"step": 47000
},
{
"epoch": 0.3768,
"grad_norm": 5.564688205718994,
"learning_rate": 6.258795180722892e-06,
"loss": 0.2634,
"step": 47100
},
{
"epoch": 0.3776,
"grad_norm": 2.6855292320251465,
"learning_rate": 6.250763052208836e-06,
"loss": 0.2509,
"step": 47200
},
{
"epoch": 0.3784,
"grad_norm": 9.584918975830078,
"learning_rate": 6.24273092369478e-06,
"loss": 0.2586,
"step": 47300
},
{
"epoch": 0.3792,
"grad_norm": 9.060691833496094,
"learning_rate": 6.2346987951807236e-06,
"loss": 0.2658,
"step": 47400
},
{
"epoch": 0.38,
"grad_norm": 3.5146710872650146,
"learning_rate": 6.2266666666666675e-06,
"loss": 0.2441,
"step": 47500
},
{
"epoch": 0.3808,
"grad_norm": 6.001715183258057,
"learning_rate": 6.218634538152611e-06,
"loss": 0.2403,
"step": 47600
},
{
"epoch": 0.3816,
"grad_norm": 3.3255221843719482,
"learning_rate": 6.2106024096385545e-06,
"loss": 0.237,
"step": 47700
},
{
"epoch": 0.3824,
"grad_norm": 9.07324504852295,
"learning_rate": 6.202570281124498e-06,
"loss": 0.2522,
"step": 47800
},
{
"epoch": 0.3832,
"grad_norm": 10.312811851501465,
"learning_rate": 6.194538152610442e-06,
"loss": 0.2356,
"step": 47900
},
{
"epoch": 0.384,
"grad_norm": 8.543242454528809,
"learning_rate": 6.186506024096386e-06,
"loss": 0.25,
"step": 48000
},
{
"epoch": 0.384,
"eval_test1_cer": 0.07607457614459631,
"eval_test1_cer_norm": 0.056354599370903063,
"eval_test1_loss": 0.19436757266521454,
"eval_test1_runtime": 2446.1615,
"eval_test1_samples_per_second": 1.022,
"eval_test1_steps_per_second": 0.256,
"eval_test1_wer": 0.1832123385521444,
"eval_test1_wer_norm": 0.12332329992109646,
"step": 48000
},
{
"epoch": 0.384,
"eval_test2_cer": 0.12698323813790272,
"eval_test2_cer_norm": 0.10085799504183453,
"eval_test2_loss": 0.3315908908843994,
"eval_test2_runtime": 2495.6856,
"eval_test2_samples_per_second": 1.002,
"eval_test2_steps_per_second": 0.25,
"eval_test2_wer": 0.25469214923323413,
"eval_test2_wer_norm": 0.1900641760256704,
"step": 48000
},
{
"epoch": 0.3848,
"grad_norm": 7.8291215896606445,
"learning_rate": 6.17847389558233e-06,
"loss": 0.2452,
"step": 48100
},
{
"epoch": 0.3856,
"grad_norm": 4.892631530761719,
"learning_rate": 6.170441767068274e-06,
"loss": 0.2568,
"step": 48200
},
{
"epoch": 0.3864,
"grad_norm": 9.617656707763672,
"learning_rate": 6.162409638554218e-06,
"loss": 0.2197,
"step": 48300
},
{
"epoch": 0.3872,
"grad_norm": 9.659303665161133,
"learning_rate": 6.154377510040162e-06,
"loss": 0.2392,
"step": 48400
},
{
"epoch": 0.388,
"grad_norm": 6.675602912902832,
"learning_rate": 6.146345381526105e-06,
"loss": 0.2209,
"step": 48500
},
{
"epoch": 0.3888,
"grad_norm": 5.2548298835754395,
"learning_rate": 6.138313253012048e-06,
"loss": 0.2642,
"step": 48600
},
{
"epoch": 0.3896,
"grad_norm": 5.428624629974365,
"learning_rate": 6.130281124497992e-06,
"loss": 0.2139,
"step": 48700
},
{
"epoch": 0.3904,
"grad_norm": 10.75515079498291,
"learning_rate": 6.122248995983936e-06,
"loss": 0.2238,
"step": 48800
},
{
"epoch": 0.3912,
"grad_norm": 9.192462921142578,
"learning_rate": 6.11421686746988e-06,
"loss": 0.2422,
"step": 48900
},
{
"epoch": 0.392,
"grad_norm": 5.55458927154541,
"learning_rate": 6.106184738955824e-06,
"loss": 0.2559,
"step": 49000
},
{
"epoch": 0.3928,
"grad_norm": 10.406610488891602,
"learning_rate": 6.098152610441768e-06,
"loss": 0.2474,
"step": 49100
},
{
"epoch": 0.3936,
"grad_norm": 6.274507999420166,
"learning_rate": 6.0901204819277116e-06,
"loss": 0.2361,
"step": 49200
},
{
"epoch": 0.3944,
"grad_norm": 10.830498695373535,
"learning_rate": 6.0820883534136555e-06,
"loss": 0.2409,
"step": 49300
},
{
"epoch": 0.3952,
"grad_norm": 11.912911415100098,
"learning_rate": 6.074056224899599e-06,
"loss": 0.2467,
"step": 49400
},
{
"epoch": 0.396,
"grad_norm": 8.484502792358398,
"learning_rate": 6.0660240963855425e-06,
"loss": 0.2411,
"step": 49500
},
{
"epoch": 0.3968,
"grad_norm": 9.443007469177246,
"learning_rate": 6.057991967871486e-06,
"loss": 0.2449,
"step": 49600
},
{
"epoch": 0.3976,
"grad_norm": 10.93896770477295,
"learning_rate": 6.04995983935743e-06,
"loss": 0.239,
"step": 49700
},
{
"epoch": 0.3984,
"grad_norm": 7.4916768074035645,
"learning_rate": 6.041927710843373e-06,
"loss": 0.2349,
"step": 49800
},
{
"epoch": 0.3992,
"grad_norm": 7.075978755950928,
"learning_rate": 6.033895582329317e-06,
"loss": 0.2418,
"step": 49900
},
{
"epoch": 0.4,
"grad_norm": 5.759922981262207,
"learning_rate": 6.025863453815261e-06,
"loss": 0.2536,
"step": 50000
},
{
"epoch": 0.4008,
"grad_norm": 10.531188011169434,
"learning_rate": 6.017831325301205e-06,
"loss": 0.2162,
"step": 50100
},
{
"epoch": 0.4016,
"grad_norm": 7.113009452819824,
"learning_rate": 6.009799196787149e-06,
"loss": 0.2373,
"step": 50200
},
{
"epoch": 0.4024,
"grad_norm": 8.859579086303711,
"learning_rate": 6.001767068273093e-06,
"loss": 0.2327,
"step": 50300
},
{
"epoch": 0.4032,
"grad_norm": 5.7207512855529785,
"learning_rate": 5.993734939759036e-06,
"loss": 0.2605,
"step": 50400
},
{
"epoch": 0.404,
"grad_norm": 5.358285427093506,
"learning_rate": 5.98570281124498e-06,
"loss": 0.2442,
"step": 50500
},
{
"epoch": 0.4048,
"grad_norm": 9.501456260681152,
"learning_rate": 5.977670682730924e-06,
"loss": 0.2333,
"step": 50600
},
{
"epoch": 0.4056,
"grad_norm": 9.682687759399414,
"learning_rate": 5.969638554216868e-06,
"loss": 0.2404,
"step": 50700
},
{
"epoch": 0.4064,
"grad_norm": 6.1207380294799805,
"learning_rate": 5.961686746987952e-06,
"loss": 0.2401,
"step": 50800
},
{
"epoch": 0.4072,
"grad_norm": 3.812893867492676,
"learning_rate": 5.953654618473896e-06,
"loss": 0.2411,
"step": 50900
},
{
"epoch": 0.408,
"grad_norm": 8.557660102844238,
"learning_rate": 5.94562248995984e-06,
"loss": 0.2486,
"step": 51000
},
{
"epoch": 0.4088,
"grad_norm": 8.239738464355469,
"learning_rate": 5.937590361445784e-06,
"loss": 0.219,
"step": 51100
},
{
"epoch": 0.4096,
"grad_norm": 8.825906753540039,
"learning_rate": 5.929558232931727e-06,
"loss": 0.2418,
"step": 51200
},
{
"epoch": 0.4104,
"grad_norm": 9.038442611694336,
"learning_rate": 5.921526104417671e-06,
"loss": 0.2495,
"step": 51300
},
{
"epoch": 0.4112,
"grad_norm": 10.421882629394531,
"learning_rate": 5.913493975903615e-06,
"loss": 0.2294,
"step": 51400
},
{
"epoch": 0.412,
"grad_norm": 5.004858493804932,
"learning_rate": 5.905461847389559e-06,
"loss": 0.2553,
"step": 51500
},
{
"epoch": 0.4128,
"grad_norm": 5.850473880767822,
"learning_rate": 5.897429718875503e-06,
"loss": 0.2443,
"step": 51600
},
{
"epoch": 0.4136,
"grad_norm": 5.483931064605713,
"learning_rate": 5.8893975903614466e-06,
"loss": 0.2849,
"step": 51700
},
{
"epoch": 0.4144,
"grad_norm": 9.20142650604248,
"learning_rate": 5.881445783132531e-06,
"loss": 0.2617,
"step": 51800
},
{
"epoch": 0.4152,
"grad_norm": 5.675454139709473,
"learning_rate": 5.873413654618474e-06,
"loss": 0.2666,
"step": 51900
},
{
"epoch": 0.416,
"grad_norm": 7.959702968597412,
"learning_rate": 5.865381526104418e-06,
"loss": 0.2286,
"step": 52000
},
{
"epoch": 0.416,
"eval_test1_cer": 0.06314885353648518,
"eval_test1_cer_norm": 0.04443057122962038,
"eval_test1_loss": 0.1907189041376114,
"eval_test1_runtime": 2490.0845,
"eval_test1_samples_per_second": 1.004,
"eval_test1_steps_per_second": 0.251,
"eval_test1_wer": 0.18128808420070555,
"eval_test1_wer_norm": 0.12083929980420234,
"step": 52000
},
{
"epoch": 0.416,
"eval_test2_cer": 0.1296011124799343,
"eval_test2_cer_norm": 0.09485396653238302,
"eval_test2_loss": 0.3308376371860504,
"eval_test2_runtime": 3341.4359,
"eval_test2_samples_per_second": 0.748,
"eval_test2_steps_per_second": 0.187,
"eval_test2_wer": 0.2722304875257496,
"eval_test2_wer_norm": 0.2069390327756131,
"step": 52000
},
{
"epoch": 0.4168,
"grad_norm": 13.770788192749023,
"learning_rate": 5.857349397590362e-06,
"loss": 0.2467,
"step": 52100
},
{
"epoch": 0.4176,
"grad_norm": 5.342738151550293,
"learning_rate": 5.849317269076306e-06,
"loss": 0.233,
"step": 52200
},
{
"epoch": 0.4184,
"grad_norm": 6.316587448120117,
"learning_rate": 5.84128514056225e-06,
"loss": 0.2513,
"step": 52300
},
{
"epoch": 0.4192,
"grad_norm": 6.252966403961182,
"learning_rate": 5.8332530120481936e-06,
"loss": 0.2495,
"step": 52400
},
{
"epoch": 0.42,
"grad_norm": 8.304986953735352,
"learning_rate": 5.8252208835341375e-06,
"loss": 0.2573,
"step": 52500
},
{
"epoch": 0.4208,
"grad_norm": 8.028290748596191,
"learning_rate": 5.817188755020081e-06,
"loss": 0.2393,
"step": 52600
},
{
"epoch": 0.4216,
"grad_norm": 10.630465507507324,
"learning_rate": 5.809156626506025e-06,
"loss": 0.2453,
"step": 52700
},
{
"epoch": 0.4224,
"grad_norm": 9.712656021118164,
"learning_rate": 5.8011244979919675e-06,
"loss": 0.2553,
"step": 52800
},
{
"epoch": 0.4232,
"grad_norm": 9.605881690979004,
"learning_rate": 5.7930923694779115e-06,
"loss": 0.2393,
"step": 52900
},
{
"epoch": 0.424,
"grad_norm": 10.953594207763672,
"learning_rate": 5.785060240963855e-06,
"loss": 0.2484,
"step": 53000
},
{
"epoch": 0.4248,
"grad_norm": 4.043101787567139,
"learning_rate": 5.777028112449799e-06,
"loss": 0.2146,
"step": 53100
},
{
"epoch": 0.4256,
"grad_norm": 8.587929725646973,
"learning_rate": 5.768995983935743e-06,
"loss": 0.2555,
"step": 53200
},
{
"epoch": 0.4264,
"grad_norm": 6.47918701171875,
"learning_rate": 5.760963855421687e-06,
"loss": 0.2277,
"step": 53300
},
{
"epoch": 0.4272,
"grad_norm": 6.032763957977295,
"learning_rate": 5.752931726907631e-06,
"loss": 0.24,
"step": 53400
},
{
"epoch": 0.428,
"grad_norm": 4.708799362182617,
"learning_rate": 5.744899598393575e-06,
"loss": 0.2262,
"step": 53500
},
{
"epoch": 0.4288,
"grad_norm": 7.626307964324951,
"learning_rate": 5.736867469879519e-06,
"loss": 0.2344,
"step": 53600
},
{
"epoch": 0.4296,
"grad_norm": 5.346840858459473,
"learning_rate": 5.728835341365462e-06,
"loss": 0.2071,
"step": 53700
},
{
"epoch": 0.4304,
"grad_norm": 6.508141040802002,
"learning_rate": 5.720803212851406e-06,
"loss": 0.2275,
"step": 53800
},
{
"epoch": 0.4312,
"grad_norm": 7.617433071136475,
"learning_rate": 5.71277108433735e-06,
"loss": 0.2359,
"step": 53900
},
{
"epoch": 0.432,
"grad_norm": 5.544378757476807,
"learning_rate": 5.704738955823294e-06,
"loss": 0.2491,
"step": 54000
},
{
"epoch": 0.4328,
"grad_norm": 5.247679233551025,
"learning_rate": 5.696706827309238e-06,
"loss": 0.2353,
"step": 54100
},
{
"epoch": 0.4336,
"grad_norm": 9.137139320373535,
"learning_rate": 5.688674698795182e-06,
"loss": 0.2397,
"step": 54200
},
{
"epoch": 0.4344,
"grad_norm": 6.74301815032959,
"learning_rate": 5.6806425702811255e-06,
"loss": 0.2602,
"step": 54300
},
{
"epoch": 0.4352,
"grad_norm": 6.35469913482666,
"learning_rate": 5.672610441767069e-06,
"loss": 0.2515,
"step": 54400
},
{
"epoch": 0.436,
"grad_norm": 10.135574340820312,
"learning_rate": 5.664578313253013e-06,
"loss": 0.2219,
"step": 54500
},
{
"epoch": 0.4368,
"grad_norm": 5.893723011016846,
"learning_rate": 5.6565461847389556e-06,
"loss": 0.2569,
"step": 54600
},
{
"epoch": 0.4376,
"grad_norm": 14.050215721130371,
"learning_rate": 5.6485140562248995e-06,
"loss": 0.2906,
"step": 54700
},
{
"epoch": 0.4384,
"grad_norm": 6.908878326416016,
"learning_rate": 5.640481927710843e-06,
"loss": 0.2444,
"step": 54800
},
{
"epoch": 0.4392,
"grad_norm": 7.57582950592041,
"learning_rate": 5.632449799196787e-06,
"loss": 0.2467,
"step": 54900
},
{
"epoch": 0.44,
"grad_norm": 11.190139770507812,
"learning_rate": 5.624417670682731e-06,
"loss": 0.2593,
"step": 55000
},
{
"epoch": 0.4408,
"grad_norm": 12.42068099975586,
"learning_rate": 5.616385542168675e-06,
"loss": 0.2453,
"step": 55100
},
{
"epoch": 0.4416,
"grad_norm": 10.345560073852539,
"learning_rate": 5.608353413654619e-06,
"loss": 0.2411,
"step": 55200
},
{
"epoch": 0.4424,
"grad_norm": 7.91195011138916,
"learning_rate": 5.600321285140563e-06,
"loss": 0.2414,
"step": 55300
},
{
"epoch": 0.4432,
"grad_norm": 6.820638179779053,
"learning_rate": 5.592289156626507e-06,
"loss": 0.2363,
"step": 55400
},
{
"epoch": 0.444,
"grad_norm": 4.268233299255371,
"learning_rate": 5.58425702811245e-06,
"loss": 0.2325,
"step": 55500
},
{
"epoch": 0.4448,
"grad_norm": 8.063896179199219,
"learning_rate": 5.576224899598394e-06,
"loss": 0.2607,
"step": 55600
},
{
"epoch": 0.4456,
"grad_norm": 3.5322296619415283,
"learning_rate": 5.568192771084338e-06,
"loss": 0.2346,
"step": 55700
},
{
"epoch": 0.4464,
"grad_norm": 9.244215965270996,
"learning_rate": 5.560160642570282e-06,
"loss": 0.2155,
"step": 55800
},
{
"epoch": 0.4472,
"grad_norm": 10.747200965881348,
"learning_rate": 5.552208835341366e-06,
"loss": 0.2351,
"step": 55900
},
{
"epoch": 0.448,
"grad_norm": 14.072000503540039,
"learning_rate": 5.54417670682731e-06,
"loss": 0.2413,
"step": 56000
},
{
"epoch": 0.448,
"eval_test1_cer": 0.0663977892299649,
"eval_test1_cer_norm": 0.0478834009652556,
"eval_test1_loss": 0.1882741004228592,
"eval_test1_runtime": 3418.3969,
"eval_test1_samples_per_second": 0.731,
"eval_test1_steps_per_second": 0.183,
"eval_test1_wer": 0.16968424735415027,
"eval_test1_wer_norm": 0.1093544522049154,
"step": 56000
},
{
"epoch": 0.448,
"eval_test2_cer": 0.13036174263635344,
"eval_test2_cer_norm": 0.09973950263402541,
"eval_test2_loss": 0.3249567449092865,
"eval_test2_runtime": 3504.8089,
"eval_test2_samples_per_second": 0.713,
"eval_test2_steps_per_second": 0.178,
"eval_test2_wer": 0.25629434653238725,
"eval_test2_wer_norm": 0.18966307586523035,
"step": 56000
},
{
"epoch": 0.4488,
"grad_norm": 8.576178550720215,
"learning_rate": 5.536144578313254e-06,
"loss": 0.2404,
"step": 56100
},
{
"epoch": 0.4496,
"grad_norm": 6.107906818389893,
"learning_rate": 5.528112449799197e-06,
"loss": 0.2217,
"step": 56200
},
{
"epoch": 0.4504,
"grad_norm": 5.073305606842041,
"learning_rate": 5.520080321285141e-06,
"loss": 0.229,
"step": 56300
},
{
"epoch": 0.4512,
"grad_norm": 5.073732376098633,
"learning_rate": 5.512048192771085e-06,
"loss": 0.2404,
"step": 56400
},
{
"epoch": 0.452,
"grad_norm": 6.29484748840332,
"learning_rate": 5.504016064257029e-06,
"loss": 0.2283,
"step": 56500
},
{
"epoch": 0.4528,
"grad_norm": 13.772027015686035,
"learning_rate": 5.495983935742973e-06,
"loss": 0.2433,
"step": 56600
},
{
"epoch": 0.4536,
"grad_norm": 6.613615989685059,
"learning_rate": 5.487951807228917e-06,
"loss": 0.2634,
"step": 56700
},
{
"epoch": 0.4544,
"grad_norm": 5.8526082038879395,
"learning_rate": 5.480000000000001e-06,
"loss": 0.2277,
"step": 56800
},
{
"epoch": 0.4552,
"grad_norm": 6.082682132720947,
"learning_rate": 5.471967871485945e-06,
"loss": 0.2372,
"step": 56900
},
{
"epoch": 0.456,
"grad_norm": 4.50082540512085,
"learning_rate": 5.463935742971888e-06,
"loss": 0.2578,
"step": 57000
},
{
"epoch": 0.4568,
"grad_norm": 5.06675910949707,
"learning_rate": 5.455903614457832e-06,
"loss": 0.2033,
"step": 57100
},
{
"epoch": 0.4576,
"grad_norm": 3.3439102172851562,
"learning_rate": 5.447871485943776e-06,
"loss": 0.2395,
"step": 57200
},
{
"epoch": 0.4584,
"grad_norm": 12.235926628112793,
"learning_rate": 5.43983935742972e-06,
"loss": 0.2335,
"step": 57300
},
{
"epoch": 0.4592,
"grad_norm": 6.169180393218994,
"learning_rate": 5.431807228915663e-06,
"loss": 0.2432,
"step": 57400
},
{
"epoch": 0.46,
"grad_norm": 3.7379937171936035,
"learning_rate": 5.423775100401607e-06,
"loss": 0.2377,
"step": 57500
},
{
"epoch": 0.4608,
"grad_norm": 6.2007060050964355,
"learning_rate": 5.4157429718875506e-06,
"loss": 0.2475,
"step": 57600
},
{
"epoch": 0.4616,
"grad_norm": 11.921826362609863,
"learning_rate": 5.4077108433734945e-06,
"loss": 0.2359,
"step": 57700
},
{
"epoch": 0.4624,
"grad_norm": 4.8144612312316895,
"learning_rate": 5.399678714859438e-06,
"loss": 0.2222,
"step": 57800
},
{
"epoch": 0.4632,
"grad_norm": 7.309458255767822,
"learning_rate": 5.3916465863453815e-06,
"loss": 0.2315,
"step": 57900
},
{
"epoch": 0.464,
"grad_norm": 6.312740325927734,
"learning_rate": 5.383614457831325e-06,
"loss": 0.2557,
"step": 58000
},
{
"epoch": 0.4648,
"grad_norm": 7.315298557281494,
"learning_rate": 5.375582329317269e-06,
"loss": 0.2673,
"step": 58100
},
{
"epoch": 0.4656,
"grad_norm": 9.65597152709961,
"learning_rate": 5.367550200803213e-06,
"loss": 0.2417,
"step": 58200
},
{
"epoch": 0.4664,
"grad_norm": 5.868946552276611,
"learning_rate": 5.359518072289157e-06,
"loss": 0.2289,
"step": 58300
},
{
"epoch": 0.4672,
"grad_norm": 7.116505146026611,
"learning_rate": 5.351485943775101e-06,
"loss": 0.2238,
"step": 58400
},
{
"epoch": 0.468,
"grad_norm": 4.75971794128418,
"learning_rate": 5.343453815261045e-06,
"loss": 0.2466,
"step": 58500
},
{
"epoch": 0.4688,
"grad_norm": 5.3648762702941895,
"learning_rate": 5.335421686746989e-06,
"loss": 0.2539,
"step": 58600
},
{
"epoch": 0.4696,
"grad_norm": 2.8886609077453613,
"learning_rate": 5.327389558232933e-06,
"loss": 0.2386,
"step": 58700
},
{
"epoch": 0.4704,
"grad_norm": 19.657976150512695,
"learning_rate": 5.319357429718875e-06,
"loss": 0.2257,
"step": 58800
},
{
"epoch": 0.4712,
"grad_norm": 6.010655879974365,
"learning_rate": 5.311325301204819e-06,
"loss": 0.2402,
"step": 58900
},
{
"epoch": 0.472,
"grad_norm": 11.614595413208008,
"learning_rate": 5.303293172690763e-06,
"loss": 0.2422,
"step": 59000
},
{
"epoch": 0.4728,
"grad_norm": 9.773826599121094,
"learning_rate": 5.295261044176707e-06,
"loss": 0.2224,
"step": 59100
},
{
"epoch": 0.4736,
"grad_norm": 3.5956804752349854,
"learning_rate": 5.287228915662651e-06,
"loss": 0.2287,
"step": 59200
},
{
"epoch": 0.4744,
"grad_norm": 5.884477615356445,
"learning_rate": 5.279196787148595e-06,
"loss": 0.2284,
"step": 59300
},
{
"epoch": 0.4752,
"grad_norm": 6.785783290863037,
"learning_rate": 5.271164658634539e-06,
"loss": 0.2486,
"step": 59400
},
{
"epoch": 0.476,
"grad_norm": 5.1648054122924805,
"learning_rate": 5.2631325301204825e-06,
"loss": 0.2352,
"step": 59500
},
{
"epoch": 0.4768,
"grad_norm": 7.0386643409729,
"learning_rate": 5.255180722891566e-06,
"loss": 0.2489,
"step": 59600
},
{
"epoch": 0.4776,
"grad_norm": 8.528487205505371,
"learning_rate": 5.24714859437751e-06,
"loss": 0.2434,
"step": 59700
},
{
"epoch": 0.4784,
"grad_norm": 10.057316780090332,
"learning_rate": 5.239116465863454e-06,
"loss": 0.232,
"step": 59800
},
{
"epoch": 0.4792,
"grad_norm": 5.452738285064697,
"learning_rate": 5.231084337349398e-06,
"loss": 0.2138,
"step": 59900
},
{
"epoch": 0.48,
"grad_norm": 12.630691528320312,
"learning_rate": 5.223052208835342e-06,
"loss": 0.2406,
"step": 60000
},
{
"epoch": 0.48,
"eval_test1_cer": 0.047455000373440887,
"eval_test1_cer_norm": 0.03161812375441208,
"eval_test1_loss": 0.18266192078590393,
"eval_test1_runtime": 3355.1591,
"eval_test1_samples_per_second": 0.745,
"eval_test1_steps_per_second": 0.186,
"eval_test1_wer": 0.1444065424647949,
"eval_test1_wer_norm": 0.08430988632046524,
"step": 60000
},
{
"epoch": 0.48,
"eval_test2_cer": 0.11545245828200246,
"eval_test2_cer_norm": 0.0919003718624109,
"eval_test2_loss": 0.32027342915534973,
"eval_test2_runtime": 3441.9039,
"eval_test2_samples_per_second": 0.726,
"eval_test2_steps_per_second": 0.182,
"eval_test2_wer": 0.24284733348592355,
"eval_test2_wer_norm": 0.17731492092596837,
"step": 60000
},
{
"epoch": 0.4808,
"grad_norm": 1.9925609827041626,
"learning_rate": 5.2150200803212856e-06,
"loss": 0.2353,
"step": 60100
},
{
"epoch": 0.4816,
"grad_norm": 5.7619476318359375,
"learning_rate": 5.2069879518072295e-06,
"loss": 0.2124,
"step": 60200
},
{
"epoch": 0.4824,
"grad_norm": 6.040704727172852,
"learning_rate": 5.198955823293173e-06,
"loss": 0.2304,
"step": 60300
},
{
"epoch": 0.4832,
"grad_norm": 14.028429985046387,
"learning_rate": 5.1909236947791165e-06,
"loss": 0.2549,
"step": 60400
},
{
"epoch": 0.484,
"grad_norm": 14.17192268371582,
"learning_rate": 5.18289156626506e-06,
"loss": 0.2215,
"step": 60500
},
{
"epoch": 0.4848,
"grad_norm": 8.68266773223877,
"learning_rate": 5.174859437751004e-06,
"loss": 0.2144,
"step": 60600
},
{
"epoch": 0.4856,
"grad_norm": 7.074737548828125,
"learning_rate": 5.166827309236948e-06,
"loss": 0.2262,
"step": 60700
},
{
"epoch": 0.4864,
"grad_norm": 7.095781326293945,
"learning_rate": 5.158795180722892e-06,
"loss": 0.2509,
"step": 60800
},
{
"epoch": 0.4872,
"grad_norm": 14.873151779174805,
"learning_rate": 5.150763052208836e-06,
"loss": 0.2417,
"step": 60900
},
{
"epoch": 0.488,
"grad_norm": 7.113455295562744,
"learning_rate": 5.14273092369478e-06,
"loss": 0.2232,
"step": 61000
},
{
"epoch": 0.4888,
"grad_norm": 4.443511009216309,
"learning_rate": 5.134698795180724e-06,
"loss": 0.2333,
"step": 61100
},
{
"epoch": 0.4896,
"grad_norm": 11.819351196289062,
"learning_rate": 5.126666666666668e-06,
"loss": 0.2252,
"step": 61200
},
{
"epoch": 0.4904,
"grad_norm": 9.132615089416504,
"learning_rate": 5.11863453815261e-06,
"loss": 0.2469,
"step": 61300
},
{
"epoch": 0.4912,
"grad_norm": 5.749680042266846,
"learning_rate": 5.110602409638554e-06,
"loss": 0.2322,
"step": 61400
},
{
"epoch": 0.492,
"grad_norm": 5.154579162597656,
"learning_rate": 5.102570281124498e-06,
"loss": 0.2293,
"step": 61500
},
{
"epoch": 0.4928,
"grad_norm": 4.018552780151367,
"learning_rate": 5.094538152610442e-06,
"loss": 0.2087,
"step": 61600
},
{
"epoch": 0.4936,
"grad_norm": 12.570889472961426,
"learning_rate": 5.086506024096386e-06,
"loss": 0.2374,
"step": 61700
},
{
"epoch": 0.4944,
"grad_norm": 4.278166770935059,
"learning_rate": 5.07847389558233e-06,
"loss": 0.2298,
"step": 61800
},
{
"epoch": 1.000696,
"grad_norm": 4.555546760559082,
"learning_rate": 5.070441767068274e-06,
"loss": 0.2077,
"step": 61900
},
{
"epoch": 1.001496,
"grad_norm": 15.487360954284668,
"learning_rate": 5.0624096385542175e-06,
"loss": 0.2044,
"step": 62000
},
{
"epoch": 1.002296,
"grad_norm": 4.297917366027832,
"learning_rate": 5.0543775100401614e-06,
"loss": 0.1977,
"step": 62100
},
{
"epoch": 1.003096,
"grad_norm": 8.327834129333496,
"learning_rate": 5.0463453815261045e-06,
"loss": 0.1749,
"step": 62200
},
{
"epoch": 1.003896,
"grad_norm": 8.811240196228027,
"learning_rate": 5.038313253012048e-06,
"loss": 0.1868,
"step": 62300
},
{
"epoch": 1.004696,
"grad_norm": 5.721954822540283,
"learning_rate": 5.030281124497992e-06,
"loss": 0.195,
"step": 62400
},
{
"epoch": 1.005496,
"grad_norm": 7.047698020935059,
"learning_rate": 5.022248995983936e-06,
"loss": 0.1605,
"step": 62500
},
{
"epoch": 1.006296,
"grad_norm": 6.009364128112793,
"learning_rate": 5.0142971887550206e-06,
"loss": 0.176,
"step": 62600
},
{
"epoch": 1.007096,
"grad_norm": 5.495716571807861,
"learning_rate": 5.0062650602409645e-06,
"loss": 0.1991,
"step": 62700
},
{
"epoch": 1.007896,
"grad_norm": 2.433659076690674,
"learning_rate": 4.9982329317269076e-06,
"loss": 0.1832,
"step": 62800
},
{
"epoch": 1.008696,
"grad_norm": 6.076290607452393,
"learning_rate": 4.9902008032128515e-06,
"loss": 0.1807,
"step": 62900
},
{
"epoch": 1.009496,
"grad_norm": 5.725612640380859,
"learning_rate": 4.982168674698795e-06,
"loss": 0.1627,
"step": 63000
},
{
"epoch": 1.010296,
"grad_norm": 4.852512836456299,
"learning_rate": 4.974136546184739e-06,
"loss": 0.1692,
"step": 63100
},
{
"epoch": 1.011096,
"grad_norm": 6.572232246398926,
"learning_rate": 4.966104417670683e-06,
"loss": 0.1616,
"step": 63200
},
{
"epoch": 1.011896,
"grad_norm": 9.408519744873047,
"learning_rate": 4.958072289156627e-06,
"loss": 0.1566,
"step": 63300
},
{
"epoch": 1.012696,
"grad_norm": 4.744565963745117,
"learning_rate": 4.950040160642571e-06,
"loss": 0.1519,
"step": 63400
},
{
"epoch": 1.013496,
"grad_norm": 13.262666702270508,
"learning_rate": 4.942008032128515e-06,
"loss": 0.1804,
"step": 63500
},
{
"epoch": 1.014296,
"grad_norm": 3.7456908226013184,
"learning_rate": 4.933975903614458e-06,
"loss": 0.1703,
"step": 63600
},
{
"epoch": 1.015096,
"grad_norm": 6.4614152908325195,
"learning_rate": 4.925943775100402e-06,
"loss": 0.1796,
"step": 63700
},
{
"epoch": 1.015896,
"grad_norm": 4.488025665283203,
"learning_rate": 4.917911646586346e-06,
"loss": 0.1678,
"step": 63800
},
{
"epoch": 1.016696,
"grad_norm": 5.777276992797852,
"learning_rate": 4.90987951807229e-06,
"loss": 0.1839,
"step": 63900
},
{
"epoch": 1.017496,
"grad_norm": 4.160243034362793,
"learning_rate": 4.901847389558234e-06,
"loss": 0.1627,
"step": 64000
},
{
"epoch": 1.017496,
"eval_test1_cer": 0.07008084995145268,
"eval_test1_cer_norm": 0.050106850433404564,
"eval_test1_loss": 0.18447566032409668,
"eval_test1_runtime": 3416.4242,
"eval_test1_samples_per_second": 0.732,
"eval_test1_steps_per_second": 0.183,
"eval_test1_wer": 0.17367853290183388,
"eval_test1_wer_norm": 0.11408866419240772,
"step": 64000
},
{
"epoch": 1.017496,
"eval_test2_cer": 0.14694627991189757,
"eval_test2_cer_norm": 0.11825515184381778,
"eval_test2_loss": 0.3210515081882477,
"eval_test2_runtime": 3532.7138,
"eval_test2_samples_per_second": 0.708,
"eval_test2_steps_per_second": 0.177,
"eval_test2_wer": 0.2882810711833371,
"eval_test2_wer_norm": 0.22352738941095576,
"step": 64000
},
{
"epoch": 1.018296,
"grad_norm": 5.366184234619141,
"learning_rate": 4.893815261044177e-06,
"loss": 0.1663,
"step": 64100
},
{
"epoch": 1.019096,
"grad_norm": 3.319096326828003,
"learning_rate": 4.885783132530121e-06,
"loss": 0.1597,
"step": 64200
},
{
"epoch": 1.019896,
"grad_norm": 2.7266533374786377,
"learning_rate": 4.877751004016065e-06,
"loss": 0.1636,
"step": 64300
},
{
"epoch": 1.020696,
"grad_norm": 5.649815082550049,
"learning_rate": 4.869718875502009e-06,
"loss": 0.1719,
"step": 64400
},
{
"epoch": 1.021496,
"grad_norm": 9.476170539855957,
"learning_rate": 4.861686746987952e-06,
"loss": 0.1612,
"step": 64500
},
{
"epoch": 1.022296,
"grad_norm": 6.443653106689453,
"learning_rate": 4.853654618473896e-06,
"loss": 0.1589,
"step": 64600
},
{
"epoch": 1.023096,
"grad_norm": 4.278164863586426,
"learning_rate": 4.8456224899598395e-06,
"loss": 0.1575,
"step": 64700
},
{
"epoch": 1.023896,
"grad_norm": 7.513009071350098,
"learning_rate": 4.837590361445783e-06,
"loss": 0.1678,
"step": 64800
},
{
"epoch": 1.024696,
"grad_norm": 7.260671615600586,
"learning_rate": 4.829558232931727e-06,
"loss": 0.1605,
"step": 64900
},
{
"epoch": 1.025496,
"grad_norm": 5.455900192260742,
"learning_rate": 4.821526104417671e-06,
"loss": 0.1555,
"step": 65000
},
{
"epoch": 1.026296,
"grad_norm": 11.086517333984375,
"learning_rate": 4.813493975903615e-06,
"loss": 0.165,
"step": 65100
},
{
"epoch": 1.027096,
"grad_norm": 4.682340145111084,
"learning_rate": 4.805461847389558e-06,
"loss": 0.1791,
"step": 65200
},
{
"epoch": 1.027896,
"grad_norm": 8.538371086120605,
"learning_rate": 4.797429718875502e-06,
"loss": 0.1656,
"step": 65300
},
{
"epoch": 1.028696,
"grad_norm": 6.110259532928467,
"learning_rate": 4.789397590361446e-06,
"loss": 0.1564,
"step": 65400
},
{
"epoch": 1.029496,
"grad_norm": 6.267505168914795,
"learning_rate": 4.78136546184739e-06,
"loss": 0.1624,
"step": 65500
},
{
"epoch": 1.030296,
"grad_norm": 7.841426372528076,
"learning_rate": 4.773333333333334e-06,
"loss": 0.1497,
"step": 65600
},
{
"epoch": 1.031096,
"grad_norm": 3.335782051086426,
"learning_rate": 4.765301204819278e-06,
"loss": 0.1684,
"step": 65700
},
{
"epoch": 1.031896,
"grad_norm": 6.293437480926514,
"learning_rate": 4.757269076305222e-06,
"loss": 0.1687,
"step": 65800
},
{
"epoch": 1.032696,
"grad_norm": 9.889056205749512,
"learning_rate": 4.749236947791165e-06,
"loss": 0.1605,
"step": 65900
},
{
"epoch": 1.033496,
"grad_norm": 7.943349838256836,
"learning_rate": 4.741204819277109e-06,
"loss": 0.1781,
"step": 66000
},
{
"epoch": 1.034296,
"grad_norm": 9.100912094116211,
"learning_rate": 4.733172690763053e-06,
"loss": 0.1789,
"step": 66100
},
{
"epoch": 1.035096,
"grad_norm": 5.020230293273926,
"learning_rate": 4.725140562248997e-06,
"loss": 0.1679,
"step": 66200
},
{
"epoch": 1.035896,
"grad_norm": 3.7011711597442627,
"learning_rate": 4.71710843373494e-06,
"loss": 0.1617,
"step": 66300
},
{
"epoch": 1.036696,
"grad_norm": 13.44151782989502,
"learning_rate": 4.709076305220884e-06,
"loss": 0.1717,
"step": 66400
},
{
"epoch": 1.037496,
"grad_norm": 5.3160319328308105,
"learning_rate": 4.7010441767068275e-06,
"loss": 0.18,
"step": 66500
},
{
"epoch": 1.0382959999999999,
"grad_norm": 9.991889953613281,
"learning_rate": 4.693092369477912e-06,
"loss": 0.1698,
"step": 66600
},
{
"epoch": 1.039096,
"grad_norm": 5.639313697814941,
"learning_rate": 4.685060240963856e-06,
"loss": 0.1595,
"step": 66700
},
{
"epoch": 1.039896,
"grad_norm": 4.916014194488525,
"learning_rate": 4.6770281124498e-06,
"loss": 0.1703,
"step": 66800
},
{
"epoch": 1.040696,
"grad_norm": 5.738508701324463,
"learning_rate": 4.668995983935744e-06,
"loss": 0.1864,
"step": 66900
},
{
"epoch": 1.041496,
"grad_norm": 8.484038352966309,
"learning_rate": 4.6609638554216875e-06,
"loss": 0.1587,
"step": 67000
},
{
"epoch": 1.042296,
"grad_norm": 6.696768283843994,
"learning_rate": 4.652931726907631e-06,
"loss": 0.1601,
"step": 67100
},
{
"epoch": 1.043096,
"grad_norm": 3.6157071590423584,
"learning_rate": 4.6448995983935745e-06,
"loss": 0.1657,
"step": 67200
},
{
"epoch": 1.043896,
"grad_norm": 6.275000095367432,
"learning_rate": 4.6368674698795184e-06,
"loss": 0.1559,
"step": 67300
},
{
"epoch": 1.044696,
"grad_norm": 11.618382453918457,
"learning_rate": 4.628835341365462e-06,
"loss": 0.1636,
"step": 67400
},
{
"epoch": 1.045496,
"grad_norm": 3.8046157360076904,
"learning_rate": 4.620803212851405e-06,
"loss": 0.165,
"step": 67500
},
{
"epoch": 1.046296,
"grad_norm": 7.296670913696289,
"learning_rate": 4.612771084337349e-06,
"loss": 0.1692,
"step": 67600
},
{
"epoch": 1.047096,
"grad_norm": 5.143111228942871,
"learning_rate": 4.604738955823293e-06,
"loss": 0.1567,
"step": 67700
},
{
"epoch": 1.047896,
"grad_norm": 6.818264961242676,
"learning_rate": 4.596706827309237e-06,
"loss": 0.1569,
"step": 67800
},
{
"epoch": 1.048696,
"grad_norm": 3.8615856170654297,
"learning_rate": 4.588674698795181e-06,
"loss": 0.1646,
"step": 67900
},
{
"epoch": 1.049496,
"grad_norm": 4.350166320800781,
"learning_rate": 4.580642570281125e-06,
"loss": 0.1562,
"step": 68000
},
{
"epoch": 1.049496,
"eval_test1_cer": 0.04518167899021585,
"eval_test1_cer_norm": 0.02964919441976613,
"eval_test1_loss": 0.18461571633815765,
"eval_test1_runtime": 3355.9439,
"eval_test1_samples_per_second": 0.745,
"eval_test1_steps_per_second": 0.186,
"eval_test1_wer": 0.1442899209889501,
"eval_test1_wer_norm": 0.08173821561121014,
"step": 68000
},
{
"epoch": 1.049496,
"eval_test2_cer": 0.08435995072236532,
"eval_test2_cer_norm": 0.06550201425472575,
"eval_test2_loss": 0.3253004252910614,
"eval_test2_runtime": 3365.7371,
"eval_test2_samples_per_second": 0.743,
"eval_test2_steps_per_second": 0.186,
"eval_test2_wer": 0.19904440375371937,
"eval_test2_wer_norm": 0.13302200320880128,
"step": 68000
},
{
"epoch": 1.050296,
"grad_norm": 7.118165016174316,
"learning_rate": 4.572610441767069e-06,
"loss": 0.1473,
"step": 68100
},
{
"epoch": 1.051096,
"grad_norm": 7.220821857452393,
"learning_rate": 4.564578313253013e-06,
"loss": 0.1617,
"step": 68200
},
{
"epoch": 1.051896,
"grad_norm": 7.148895740509033,
"learning_rate": 4.556546184738957e-06,
"loss": 0.1629,
"step": 68300
},
{
"epoch": 1.052696,
"grad_norm": 3.8178248405456543,
"learning_rate": 4.5485140562249e-06,
"loss": 0.181,
"step": 68400
},
{
"epoch": 1.053496,
"grad_norm": 6.640100479125977,
"learning_rate": 4.540481927710844e-06,
"loss": 0.1545,
"step": 68500
},
{
"epoch": 1.054296,
"grad_norm": 6.102357387542725,
"learning_rate": 4.532530120481928e-06,
"loss": 0.1716,
"step": 68600
},
{
"epoch": 1.055096,
"grad_norm": 5.749415397644043,
"learning_rate": 4.524497991967872e-06,
"loss": 0.1419,
"step": 68700
},
{
"epoch": 1.055896,
"grad_norm": 5.71680212020874,
"learning_rate": 4.516465863453816e-06,
"loss": 0.1487,
"step": 68800
},
{
"epoch": 1.056696,
"grad_norm": 5.209336757659912,
"learning_rate": 4.50843373493976e-06,
"loss": 0.1722,
"step": 68900
},
{
"epoch": 1.057496,
"grad_norm": 6.00562047958374,
"learning_rate": 4.500401606425703e-06,
"loss": 0.1659,
"step": 69000
},
{
"epoch": 1.058296,
"grad_norm": 4.806443214416504,
"learning_rate": 4.492369477911647e-06,
"loss": 0.1499,
"step": 69100
},
{
"epoch": 1.059096,
"grad_norm": 7.675013065338135,
"learning_rate": 4.484337349397591e-06,
"loss": 0.1624,
"step": 69200
},
{
"epoch": 1.059896,
"grad_norm": 5.567787170410156,
"learning_rate": 4.476305220883535e-06,
"loss": 0.1572,
"step": 69300
},
{
"epoch": 1.060696,
"grad_norm": 20.12483787536621,
"learning_rate": 4.468273092369479e-06,
"loss": 0.169,
"step": 69400
},
{
"epoch": 1.061496,
"grad_norm": 11.066375732421875,
"learning_rate": 4.4602409638554225e-06,
"loss": 0.1618,
"step": 69500
},
{
"epoch": 1.062296,
"grad_norm": 6.3260579109191895,
"learning_rate": 4.452208835341366e-06,
"loss": 0.1465,
"step": 69600
},
{
"epoch": 1.063096,
"grad_norm": 5.906326770782471,
"learning_rate": 4.4441767068273095e-06,
"loss": 0.1532,
"step": 69700
},
{
"epoch": 1.063896,
"grad_norm": 4.177192211151123,
"learning_rate": 4.4361445783132534e-06,
"loss": 0.1624,
"step": 69800
},
{
"epoch": 1.064696,
"grad_norm": 4.535795211791992,
"learning_rate": 4.428112449799197e-06,
"loss": 0.1637,
"step": 69900
},
{
"epoch": 1.065496,
"grad_norm": 8.556774139404297,
"learning_rate": 4.420080321285141e-06,
"loss": 0.1512,
"step": 70000
},
{
"epoch": 1.066296,
"grad_norm": 5.175086975097656,
"learning_rate": 4.412048192771084e-06,
"loss": 0.1487,
"step": 70100
},
{
"epoch": 1.067096,
"grad_norm": 5.598880767822266,
"learning_rate": 4.404016064257028e-06,
"loss": 0.1604,
"step": 70200
},
{
"epoch": 1.067896,
"grad_norm": 4.312199115753174,
"learning_rate": 4.395983935742972e-06,
"loss": 0.1752,
"step": 70300
},
{
"epoch": 1.068696,
"grad_norm": 4.8355255126953125,
"learning_rate": 4.387951807228916e-06,
"loss": 0.1375,
"step": 70400
},
{
"epoch": 1.069496,
"grad_norm": 0.6319503784179688,
"learning_rate": 4.379919678714859e-06,
"loss": 0.1518,
"step": 70500
},
{
"epoch": 1.070296,
"grad_norm": 7.824057579040527,
"learning_rate": 4.371887550200803e-06,
"loss": 0.171,
"step": 70600
},
{
"epoch": 1.071096,
"grad_norm": 6.540349960327148,
"learning_rate": 4.363855421686747e-06,
"loss": 0.146,
"step": 70700
},
{
"epoch": 1.071896,
"grad_norm": 11.684809684753418,
"learning_rate": 4.355823293172691e-06,
"loss": 0.1504,
"step": 70800
},
{
"epoch": 1.072696,
"grad_norm": 9.533754348754883,
"learning_rate": 4.347791164658635e-06,
"loss": 0.1548,
"step": 70900
},
{
"epoch": 1.073496,
"grad_norm": 6.979349136352539,
"learning_rate": 4.339759036144579e-06,
"loss": 0.1608,
"step": 71000
},
{
"epoch": 1.074296,
"grad_norm": 6.119946002960205,
"learning_rate": 4.331807228915663e-06,
"loss": 0.1642,
"step": 71100
},
{
"epoch": 1.075096,
"grad_norm": 5.234997272491455,
"learning_rate": 4.323775100401606e-06,
"loss": 0.159,
"step": 71200
},
{
"epoch": 1.075896,
"grad_norm": 11.375402450561523,
"learning_rate": 4.31574297188755e-06,
"loss": 0.1598,
"step": 71300
},
{
"epoch": 1.076696,
"grad_norm": 6.117208003997803,
"learning_rate": 4.307710843373494e-06,
"loss": 0.1581,
"step": 71400
},
{
"epoch": 1.077496,
"grad_norm": 5.760785102844238,
"learning_rate": 4.299678714859438e-06,
"loss": 0.1581,
"step": 71500
},
{
"epoch": 1.078296,
"grad_norm": 12.158953666687012,
"learning_rate": 4.291646586345382e-06,
"loss": 0.1645,
"step": 71600
},
{
"epoch": 1.079096,
"grad_norm": 3.9681832790374756,
"learning_rate": 4.283614457831326e-06,
"loss": 0.1664,
"step": 71700
},
{
"epoch": 1.079896,
"grad_norm": 4.026464939117432,
"learning_rate": 4.27558232931727e-06,
"loss": 0.1478,
"step": 71800
},
{
"epoch": 1.080696,
"grad_norm": 9.842916488647461,
"learning_rate": 4.267550200803214e-06,
"loss": 0.1516,
"step": 71900
},
{
"epoch": 1.081496,
"grad_norm": 6.5888776779174805,
"learning_rate": 4.2595180722891575e-06,
"loss": 0.1743,
"step": 72000
},
{
"epoch": 1.081496,
"eval_test1_cer": 0.056235529165733066,
"eval_test1_cer_norm": 0.03523423055682282,
"eval_test1_loss": 0.18351316452026367,
"eval_test1_runtime": 3389.512,
"eval_test1_samples_per_second": 0.738,
"eval_test1_steps_per_second": 0.184,
"eval_test1_wer": 0.15283244409458002,
"eval_test1_wer_norm": 0.09255092200239633,
"step": 72000
},
{
"epoch": 1.081496,
"eval_test2_cer": 0.10279239929816703,
"eval_test2_cer_norm": 0.0767353579175705,
"eval_test2_loss": 0.32321926951408386,
"eval_test2_runtime": 3400.9588,
"eval_test2_samples_per_second": 0.735,
"eval_test2_steps_per_second": 0.184,
"eval_test2_wer": 0.21498054474708173,
"eval_test2_wer_norm": 0.14843570937428374,
"step": 72000
},
{
"epoch": 1.082296,
"grad_norm": 11.681933403015137,
"learning_rate": 4.251485943775101e-06,
"loss": 0.1564,
"step": 72100
},
{
"epoch": 1.083096,
"grad_norm": 5.162091255187988,
"learning_rate": 4.2434538152610445e-06,
"loss": 0.1469,
"step": 72200
},
{
"epoch": 1.083896,
"grad_norm": 7.147501468658447,
"learning_rate": 4.2354216867469884e-06,
"loss": 0.1663,
"step": 72300
},
{
"epoch": 1.084696,
"grad_norm": 7.40344762802124,
"learning_rate": 4.227389558232932e-06,
"loss": 0.1284,
"step": 72400
},
{
"epoch": 1.085496,
"grad_norm": 3.5983896255493164,
"learning_rate": 4.219357429718876e-06,
"loss": 0.1558,
"step": 72500
},
{
"epoch": 1.086296,
"grad_norm": 4.397582054138184,
"learning_rate": 4.211325301204819e-06,
"loss": 0.1599,
"step": 72600
},
{
"epoch": 1.087096,
"grad_norm": 2.2897703647613525,
"learning_rate": 4.203293172690763e-06,
"loss": 0.1611,
"step": 72700
},
{
"epoch": 1.087896,
"grad_norm": 4.20655632019043,
"learning_rate": 4.195261044176707e-06,
"loss": 0.1597,
"step": 72800
},
{
"epoch": 1.088696,
"grad_norm": 4.474186420440674,
"learning_rate": 4.187228915662651e-06,
"loss": 0.145,
"step": 72900
},
{
"epoch": 1.089496,
"grad_norm": 4.5594401359558105,
"learning_rate": 4.179196787148594e-06,
"loss": 0.1545,
"step": 73000
},
{
"epoch": 1.090296,
"grad_norm": 3.502408266067505,
"learning_rate": 4.171164658634538e-06,
"loss": 0.1648,
"step": 73100
},
{
"epoch": 1.091096,
"grad_norm": 2.6792593002319336,
"learning_rate": 4.163132530120482e-06,
"loss": 0.1491,
"step": 73200
},
{
"epoch": 1.091896,
"grad_norm": 6.028800010681152,
"learning_rate": 4.155100401606426e-06,
"loss": 0.1544,
"step": 73300
},
{
"epoch": 1.0926960000000001,
"grad_norm": 5.256026268005371,
"learning_rate": 4.14706827309237e-06,
"loss": 0.1526,
"step": 73400
},
{
"epoch": 1.093496,
"grad_norm": 5.323427677154541,
"learning_rate": 4.139036144578314e-06,
"loss": 0.1412,
"step": 73500
},
{
"epoch": 1.094296,
"grad_norm": 3.4750475883483887,
"learning_rate": 4.131004016064257e-06,
"loss": 0.1372,
"step": 73600
},
{
"epoch": 1.095096,
"grad_norm": 2.8323254585266113,
"learning_rate": 4.122971887550201e-06,
"loss": 0.156,
"step": 73700
},
{
"epoch": 1.095896,
"grad_norm": 11.332598686218262,
"learning_rate": 4.114939759036145e-06,
"loss": 0.1443,
"step": 73800
},
{
"epoch": 1.0966960000000001,
"grad_norm": 3.0766139030456543,
"learning_rate": 4.106907630522089e-06,
"loss": 0.1513,
"step": 73900
},
{
"epoch": 1.097496,
"grad_norm": 8.13890266418457,
"learning_rate": 4.0988755020080325e-06,
"loss": 0.1441,
"step": 74000
},
{
"epoch": 1.098296,
"grad_norm": 6.174729824066162,
"learning_rate": 4.0908433734939765e-06,
"loss": 0.1465,
"step": 74100
},
{
"epoch": 1.099096,
"grad_norm": 7.782393455505371,
"learning_rate": 4.08281124497992e-06,
"loss": 0.1525,
"step": 74200
},
{
"epoch": 1.099896,
"grad_norm": 4.48836088180542,
"learning_rate": 4.074779116465864e-06,
"loss": 0.1419,
"step": 74300
},
{
"epoch": 1.100696,
"grad_norm": 9.527304649353027,
"learning_rate": 4.066746987951807e-06,
"loss": 0.1471,
"step": 74400
},
{
"epoch": 1.101496,
"grad_norm": 5.195642471313477,
"learning_rate": 4.058714859437751e-06,
"loss": 0.1454,
"step": 74500
},
{
"epoch": 1.102296,
"grad_norm": 4.529500961303711,
"learning_rate": 4.050682730923695e-06,
"loss": 0.1582,
"step": 74600
},
{
"epoch": 1.103096,
"grad_norm": 4.841000080108643,
"learning_rate": 4.042650602409639e-06,
"loss": 0.1521,
"step": 74700
},
{
"epoch": 1.103896,
"grad_norm": 4.943587303161621,
"learning_rate": 4.034618473895583e-06,
"loss": 0.148,
"step": 74800
},
{
"epoch": 1.104696,
"grad_norm": 7.331528186798096,
"learning_rate": 4.026586345381526e-06,
"loss": 0.1469,
"step": 74900
},
{
"epoch": 1.105496,
"grad_norm": 8.253586769104004,
"learning_rate": 4.01855421686747e-06,
"loss": 0.1736,
"step": 75000
},
{
"epoch": 1.106296,
"grad_norm": 6.348284721374512,
"learning_rate": 4.010602409638554e-06,
"loss": 0.1398,
"step": 75100
},
{
"epoch": 1.107096,
"grad_norm": 2.285827398300171,
"learning_rate": 4.002570281124498e-06,
"loss": 0.146,
"step": 75200
},
{
"epoch": 1.107896,
"grad_norm": 7.164559364318848,
"learning_rate": 3.994538152610442e-06,
"loss": 0.1564,
"step": 75300
},
{
"epoch": 1.108696,
"grad_norm": 2.311056137084961,
"learning_rate": 3.986506024096386e-06,
"loss": 0.1509,
"step": 75400
},
{
"epoch": 1.109496,
"grad_norm": 3.1986663341522217,
"learning_rate": 3.97847389558233e-06,
"loss": 0.1521,
"step": 75500
},
{
"epoch": 1.110296,
"grad_norm": 10.752972602844238,
"learning_rate": 3.970441767068273e-06,
"loss": 0.1321,
"step": 75600
},
{
"epoch": 1.111096,
"grad_norm": 5.435462474822998,
"learning_rate": 3.962409638554217e-06,
"loss": 0.1413,
"step": 75700
},
{
"epoch": 1.111896,
"grad_norm": 3.7016944885253906,
"learning_rate": 3.954457831325301e-06,
"loss": 0.1661,
"step": 75800
},
{
"epoch": 1.112696,
"grad_norm": 9.687053680419922,
"learning_rate": 3.946425702811245e-06,
"loss": 0.1516,
"step": 75900
},
{
"epoch": 1.113496,
"grad_norm": 2.699572801589966,
"learning_rate": 3.938393574297189e-06,
"loss": 0.153,
"step": 76000
},
{
"epoch": 1.113496,
"eval_test1_cer": 0.07078571962058407,
"eval_test1_cer_norm": 0.04619780536413187,
"eval_test1_loss": 0.18133017420768738,
"eval_test1_runtime": 3430.9909,
"eval_test1_samples_per_second": 0.729,
"eval_test1_steps_per_second": 0.182,
"eval_test1_wer": 0.16875127554739205,
"eval_test1_wer_norm": 0.11002659341301615,
"step": 76000
},
{
"epoch": 1.113496,
"eval_test2_cer": 0.13182233919438532,
"eval_test2_cer_norm": 0.09897931515339324,
"eval_test2_loss": 0.32256972789764404,
"eval_test2_runtime": 3529.0223,
"eval_test2_samples_per_second": 0.708,
"eval_test2_steps_per_second": 0.177,
"eval_test2_wer": 0.26012817578393227,
"eval_test2_wer_norm": 0.19547902819161128,
"step": 76000
},
{
"epoch": 1.114296,
"grad_norm": 4.699991703033447,
"learning_rate": 3.930361445783133e-06,
"loss": 0.1306,
"step": 76100
},
{
"epoch": 1.115096,
"grad_norm": 5.847876071929932,
"learning_rate": 3.922329317269077e-06,
"loss": 0.159,
"step": 76200
},
{
"epoch": 1.115896,
"grad_norm": 14.771410942077637,
"learning_rate": 3.91429718875502e-06,
"loss": 0.1593,
"step": 76300
},
{
"epoch": 1.116696,
"grad_norm": 3.190328598022461,
"learning_rate": 3.906265060240964e-06,
"loss": 0.1513,
"step": 76400
},
{
"epoch": 1.117496,
"grad_norm": 3.2370731830596924,
"learning_rate": 3.898232931726908e-06,
"loss": 0.1524,
"step": 76500
},
{
"epoch": 1.118296,
"grad_norm": 3.932393789291382,
"learning_rate": 3.890200803212852e-06,
"loss": 0.129,
"step": 76600
},
{
"epoch": 1.119096,
"grad_norm": 3.998230457305908,
"learning_rate": 3.882168674698796e-06,
"loss": 0.1534,
"step": 76700
},
{
"epoch": 1.119896,
"grad_norm": 5.405791282653809,
"learning_rate": 3.874136546184739e-06,
"loss": 0.1455,
"step": 76800
},
{
"epoch": 1.120696,
"grad_norm": 5.66414737701416,
"learning_rate": 3.866104417670683e-06,
"loss": 0.1321,
"step": 76900
},
{
"epoch": 1.121496,
"grad_norm": 2.428980827331543,
"learning_rate": 3.858072289156627e-06,
"loss": 0.1251,
"step": 77000
},
{
"epoch": 1.122296,
"grad_norm": 8.633134841918945,
"learning_rate": 3.850040160642571e-06,
"loss": 0.1581,
"step": 77100
},
{
"epoch": 1.123096,
"grad_norm": 6.84693717956543,
"learning_rate": 3.8420080321285145e-06,
"loss": 0.1516,
"step": 77200
},
{
"epoch": 1.123896,
"grad_norm": 1.603908896446228,
"learning_rate": 3.8339759036144584e-06,
"loss": 0.1614,
"step": 77300
},
{
"epoch": 1.124696,
"grad_norm": 12.588580131530762,
"learning_rate": 3.8259437751004015e-06,
"loss": 0.1423,
"step": 77400
},
{
"epoch": 1.125496,
"grad_norm": 8.18181324005127,
"learning_rate": 3.8179116465863454e-06,
"loss": 0.1409,
"step": 77500
},
{
"epoch": 1.126296,
"grad_norm": 4.0206217765808105,
"learning_rate": 3.8098795180722898e-06,
"loss": 0.1407,
"step": 77600
},
{
"epoch": 1.1270959999999999,
"grad_norm": 3.093519449234009,
"learning_rate": 3.8018473895582333e-06,
"loss": 0.128,
"step": 77700
},
{
"epoch": 1.127896,
"grad_norm": 5.9990925788879395,
"learning_rate": 3.7938152610441768e-06,
"loss": 0.138,
"step": 77800
},
{
"epoch": 1.128696,
"grad_norm": 5.512988567352295,
"learning_rate": 3.7857831325301207e-06,
"loss": 0.1366,
"step": 77900
},
{
"epoch": 1.129496,
"grad_norm": 2.798251152038574,
"learning_rate": 3.7777510040160646e-06,
"loss": 0.141,
"step": 78000
},
{
"epoch": 1.130296,
"grad_norm": 4.432553768157959,
"learning_rate": 3.769718875502008e-06,
"loss": 0.1415,
"step": 78100
},
{
"epoch": 1.1310959999999999,
"grad_norm": 4.363057613372803,
"learning_rate": 3.761686746987952e-06,
"loss": 0.1516,
"step": 78200
},
{
"epoch": 1.131896,
"grad_norm": 8.875186920166016,
"learning_rate": 3.753654618473896e-06,
"loss": 0.1578,
"step": 78300
},
{
"epoch": 1.132696,
"grad_norm": 5.948533535003662,
"learning_rate": 3.74562248995984e-06,
"loss": 0.1436,
"step": 78400
},
{
"epoch": 1.133496,
"grad_norm": 3.6585094928741455,
"learning_rate": 3.737590361445784e-06,
"loss": 0.1452,
"step": 78500
},
{
"epoch": 1.134296,
"grad_norm": 4.998322486877441,
"learning_rate": 3.729558232931727e-06,
"loss": 0.1395,
"step": 78600
},
{
"epoch": 1.1350959999999999,
"grad_norm": 2.1381640434265137,
"learning_rate": 3.7215261044176708e-06,
"loss": 0.1386,
"step": 78700
},
{
"epoch": 1.135896,
"grad_norm": 3.784778594970703,
"learning_rate": 3.7134939759036147e-06,
"loss": 0.1494,
"step": 78800
},
{
"epoch": 1.136696,
"grad_norm": 12.97223949432373,
"learning_rate": 3.7054618473895586e-06,
"loss": 0.1389,
"step": 78900
},
{
"epoch": 1.137496,
"grad_norm": 2.718700647354126,
"learning_rate": 3.697429718875502e-06,
"loss": 0.1488,
"step": 79000
},
{
"epoch": 1.138296,
"grad_norm": 5.164200782775879,
"learning_rate": 3.689397590361446e-06,
"loss": 0.1554,
"step": 79100
},
{
"epoch": 1.1390959999999999,
"grad_norm": 5.673985004425049,
"learning_rate": 3.68136546184739e-06,
"loss": 0.136,
"step": 79200
},
{
"epoch": 1.139896,
"grad_norm": 2.9805924892425537,
"learning_rate": 3.673333333333334e-06,
"loss": 0.1394,
"step": 79300
},
{
"epoch": 1.140696,
"grad_norm": 7.971452713012695,
"learning_rate": 3.665301204819278e-06,
"loss": 0.1515,
"step": 79400
},
{
"epoch": 1.141496,
"grad_norm": 3.0903499126434326,
"learning_rate": 3.657269076305221e-06,
"loss": 0.1378,
"step": 79500
},
{
"epoch": 1.142296,
"grad_norm": 4.741939067840576,
"learning_rate": 3.649236947791165e-06,
"loss": 0.1472,
"step": 79600
},
{
"epoch": 1.143096,
"grad_norm": 5.633648872375488,
"learning_rate": 3.6412048192771087e-06,
"loss": 0.1573,
"step": 79700
},
{
"epoch": 1.143896,
"grad_norm": 3.337674379348755,
"learning_rate": 3.633253012048193e-06,
"loss": 0.1531,
"step": 79800
},
{
"epoch": 1.144696,
"grad_norm": 15.155500411987305,
"learning_rate": 3.625220883534137e-06,
"loss": 0.1438,
"step": 79900
},
{
"epoch": 1.145496,
"grad_norm": 7.186131477355957,
"learning_rate": 3.617188755020081e-06,
"loss": 0.1315,
"step": 80000
},
{
"epoch": 1.145496,
"eval_test1_cer": 0.050181118828889384,
"eval_test1_cer_norm": 0.03300117655533412,
"eval_test1_loss": 0.17834880948066711,
"eval_test1_runtime": 3408.2771,
"eval_test1_samples_per_second": 0.734,
"eval_test1_steps_per_second": 0.183,
"eval_test1_wer": 0.1458643109128546,
"eval_test1_wer_norm": 0.08547882755194483,
"step": 80000
},
{
"epoch": 1.145496,
"eval_test2_cer": 0.10706219434800462,
"eval_test2_cer_norm": 0.08262802138208862,
"eval_test2_loss": 0.3213089108467102,
"eval_test2_runtime": 3477.8704,
"eval_test2_samples_per_second": 0.719,
"eval_test2_steps_per_second": 0.18,
"eval_test2_wer": 0.23108834973678188,
"eval_test2_wer_norm": 0.16551111620444647,
"step": 80000
},
{
"epoch": 1.146296,
"grad_norm": 5.269469738006592,
"learning_rate": 3.6091566265060248e-06,
"loss": 0.1244,
"step": 80100
},
{
"epoch": 1.147096,
"grad_norm": 5.666225433349609,
"learning_rate": 3.6012048192771087e-06,
"loss": 0.13,
"step": 80200
},
{
"epoch": 1.147896,
"grad_norm": 5.9655375480651855,
"learning_rate": 3.5931726907630526e-06,
"loss": 0.1464,
"step": 80300
},
{
"epoch": 1.148696,
"grad_norm": 2.792882204055786,
"learning_rate": 3.5851405622489965e-06,
"loss": 0.1482,
"step": 80400
},
{
"epoch": 1.149496,
"grad_norm": 2.3720922470092773,
"learning_rate": 3.57710843373494e-06,
"loss": 0.1375,
"step": 80500
},
{
"epoch": 1.150296,
"grad_norm": 3.8564910888671875,
"learning_rate": 3.569076305220884e-06,
"loss": 0.1291,
"step": 80600
},
{
"epoch": 1.151096,
"grad_norm": 5.648325443267822,
"learning_rate": 3.561044176706828e-06,
"loss": 0.1341,
"step": 80700
},
{
"epoch": 1.151896,
"grad_norm": 8.060567855834961,
"learning_rate": 3.5530120481927718e-06,
"loss": 0.1285,
"step": 80800
},
{
"epoch": 1.152696,
"grad_norm": 2.7208454608917236,
"learning_rate": 3.544979919678715e-06,
"loss": 0.1413,
"step": 80900
},
{
"epoch": 1.153496,
"grad_norm": 5.355711460113525,
"learning_rate": 3.5369477911646588e-06,
"loss": 0.1259,
"step": 81000
},
{
"epoch": 1.154296,
"grad_norm": 7.103275775909424,
"learning_rate": 3.5289156626506027e-06,
"loss": 0.149,
"step": 81100
},
{
"epoch": 1.155096,
"grad_norm": 3.506634473800659,
"learning_rate": 3.5208835341365466e-06,
"loss": 0.1414,
"step": 81200
},
{
"epoch": 1.155896,
"grad_norm": 5.049018383026123,
"learning_rate": 3.5128514056224905e-06,
"loss": 0.1404,
"step": 81300
},
{
"epoch": 1.156696,
"grad_norm": 4.586015701293945,
"learning_rate": 3.504819277108434e-06,
"loss": 0.1446,
"step": 81400
},
{
"epoch": 1.157496,
"grad_norm": 4.436155796051025,
"learning_rate": 3.496787148594378e-06,
"loss": 0.1406,
"step": 81500
},
{
"epoch": 1.158296,
"grad_norm": 2.5560381412506104,
"learning_rate": 3.4887550200803214e-06,
"loss": 0.1384,
"step": 81600
},
{
"epoch": 1.159096,
"grad_norm": 5.3852057456970215,
"learning_rate": 3.4807228915662654e-06,
"loss": 0.1412,
"step": 81700
},
{
"epoch": 1.159896,
"grad_norm": 5.565433502197266,
"learning_rate": 3.472690763052209e-06,
"loss": 0.1441,
"step": 81800
},
{
"epoch": 1.160696,
"grad_norm": 10.139933586120605,
"learning_rate": 3.4646586345381528e-06,
"loss": 0.1387,
"step": 81900
},
{
"epoch": 1.161496,
"grad_norm": 10.769366264343262,
"learning_rate": 3.4566265060240967e-06,
"loss": 0.1424,
"step": 82000
},
{
"epoch": 1.162296,
"grad_norm": 7.583619594573975,
"learning_rate": 3.4485943775100406e-06,
"loss": 0.1476,
"step": 82100
},
{
"epoch": 1.163096,
"grad_norm": 3.2200429439544678,
"learning_rate": 3.4405622489959845e-06,
"loss": 0.1391,
"step": 82200
},
{
"epoch": 1.163896,
"grad_norm": 4.72890567779541,
"learning_rate": 3.4325301204819276e-06,
"loss": 0.1371,
"step": 82300
},
{
"epoch": 1.164696,
"grad_norm": 3.9485466480255127,
"learning_rate": 3.4244979919678715e-06,
"loss": 0.1349,
"step": 82400
},
{
"epoch": 1.165496,
"grad_norm": 7.35235071182251,
"learning_rate": 3.4164658634538154e-06,
"loss": 0.1419,
"step": 82500
},
{
"epoch": 1.166296,
"grad_norm": 7.841335773468018,
"learning_rate": 3.4084337349397594e-06,
"loss": 0.1459,
"step": 82600
},
{
"epoch": 1.167096,
"grad_norm": 3.344604015350342,
"learning_rate": 3.400401606425703e-06,
"loss": 0.1198,
"step": 82700
},
{
"epoch": 1.167896,
"grad_norm": 4.77374792098999,
"learning_rate": 3.3923694779116468e-06,
"loss": 0.1531,
"step": 82800
},
{
"epoch": 1.168696,
"grad_norm": 6.152327537536621,
"learning_rate": 3.3843373493975907e-06,
"loss": 0.1303,
"step": 82900
},
{
"epoch": 1.169496,
"grad_norm": 5.056685924530029,
"learning_rate": 3.3763052208835346e-06,
"loss": 0.1479,
"step": 83000
},
{
"epoch": 1.170296,
"grad_norm": 7.288972854614258,
"learning_rate": 3.3682730923694785e-06,
"loss": 0.1502,
"step": 83100
},
{
"epoch": 1.171096,
"grad_norm": 6.630878448486328,
"learning_rate": 3.3602409638554216e-06,
"loss": 0.1435,
"step": 83200
},
{
"epoch": 1.171896,
"grad_norm": 6.784937381744385,
"learning_rate": 3.3522088353413655e-06,
"loss": 0.1445,
"step": 83300
},
{
"epoch": 1.172696,
"grad_norm": 5.813895225524902,
"learning_rate": 3.3441767068273095e-06,
"loss": 0.1362,
"step": 83400
},
{
"epoch": 1.173496,
"grad_norm": 6.4298415184021,
"learning_rate": 3.3361445783132534e-06,
"loss": 0.138,
"step": 83500
},
{
"epoch": 1.174296,
"grad_norm": 3.369779586791992,
"learning_rate": 3.328112449799197e-06,
"loss": 0.1251,
"step": 83600
},
{
"epoch": 1.175096,
"grad_norm": 2.4179258346557617,
"learning_rate": 3.320080321285141e-06,
"loss": 0.1491,
"step": 83700
},
{
"epoch": 1.175896,
"grad_norm": 6.803851127624512,
"learning_rate": 3.3120481927710847e-06,
"loss": 0.1209,
"step": 83800
},
{
"epoch": 1.176696,
"grad_norm": 5.81317138671875,
"learning_rate": 3.3040160642570286e-06,
"loss": 0.1463,
"step": 83900
},
{
"epoch": 1.177496,
"grad_norm": 3.342420816421509,
"learning_rate": 3.2959839357429726e-06,
"loss": 0.1108,
"step": 84000
},
{
"epoch": 1.177496,
"eval_test1_cer": 0.04321177832549108,
"eval_test1_cer_norm": 0.028405407352270273,
"eval_test1_loss": 0.17878110706806183,
"eval_test1_runtime": 3382.137,
"eval_test1_samples_per_second": 0.739,
"eval_test1_steps_per_second": 0.185,
"eval_test1_wer": 0.13557246566955306,
"eval_test1_wer_norm": 0.07484146234548059,
"step": 84000
},
{
"epoch": 1.177496,
"eval_test2_cer": 0.10907343114197185,
"eval_test2_cer_norm": 0.08453091106290672,
"eval_test2_loss": 0.32212311029434204,
"eval_test2_runtime": 3480.5629,
"eval_test2_samples_per_second": 0.718,
"eval_test2_steps_per_second": 0.18,
"eval_test2_wer": 0.2316891737239643,
"eval_test2_wer_norm": 0.16588356635342655,
"step": 84000
},
{
"epoch": 1.178296,
"grad_norm": 5.4935503005981445,
"learning_rate": 3.2879518072289156e-06,
"loss": 0.1458,
"step": 84100
},
{
"epoch": 1.179096,
"grad_norm": 7.605450630187988,
"learning_rate": 3.2799196787148595e-06,
"loss": 0.1377,
"step": 84200
},
{
"epoch": 1.179896,
"grad_norm": 6.349878311157227,
"learning_rate": 3.2719678714859443e-06,
"loss": 0.1365,
"step": 84300
},
{
"epoch": 1.180696,
"grad_norm": 6.173891544342041,
"learning_rate": 3.2639357429718878e-06,
"loss": 0.1394,
"step": 84400
},
{
"epoch": 1.181496,
"grad_norm": 1.916458249092102,
"learning_rate": 3.2559036144578317e-06,
"loss": 0.1343,
"step": 84500
},
{
"epoch": 1.182296,
"grad_norm": 4.605401039123535,
"learning_rate": 3.2478714859437756e-06,
"loss": 0.1438,
"step": 84600
},
{
"epoch": 1.183096,
"grad_norm": 10.40539264678955,
"learning_rate": 3.2398393574297195e-06,
"loss": 0.1426,
"step": 84700
},
{
"epoch": 1.183896,
"grad_norm": 5.316925525665283,
"learning_rate": 3.2318072289156626e-06,
"loss": 0.1312,
"step": 84800
},
{
"epoch": 1.184696,
"grad_norm": 3.3205056190490723,
"learning_rate": 3.2237751004016065e-06,
"loss": 0.1339,
"step": 84900
},
{
"epoch": 1.185496,
"grad_norm": 4.223670959472656,
"learning_rate": 3.2157429718875504e-06,
"loss": 0.1477,
"step": 85000
},
{
"epoch": 1.186296,
"grad_norm": 2.418142318725586,
"learning_rate": 3.2077108433734944e-06,
"loss": 0.13,
"step": 85100
},
{
"epoch": 1.187096,
"grad_norm": 5.908360958099365,
"learning_rate": 3.1996787148594383e-06,
"loss": 0.1351,
"step": 85200
},
{
"epoch": 1.187896,
"grad_norm": 3.79148268699646,
"learning_rate": 3.1916465863453818e-06,
"loss": 0.1193,
"step": 85300
},
{
"epoch": 1.188696,
"grad_norm": 4.751871109008789,
"learning_rate": 3.1836144578313257e-06,
"loss": 0.1357,
"step": 85400
},
{
"epoch": 1.189496,
"grad_norm": 5.901039123535156,
"learning_rate": 3.175582329317269e-06,
"loss": 0.1214,
"step": 85500
},
{
"epoch": 1.190296,
"grad_norm": 4.6331305503845215,
"learning_rate": 3.167550200803213e-06,
"loss": 0.1593,
"step": 85600
},
{
"epoch": 1.191096,
"grad_norm": 5.742949962615967,
"learning_rate": 3.1595180722891566e-06,
"loss": 0.1321,
"step": 85700
},
{
"epoch": 1.191896,
"grad_norm": 3.329071283340454,
"learning_rate": 3.1514859437751005e-06,
"loss": 0.1514,
"step": 85800
},
{
"epoch": 1.192696,
"grad_norm": 6.989814758300781,
"learning_rate": 3.1434538152610445e-06,
"loss": 0.135,
"step": 85900
},
{
"epoch": 1.1934960000000001,
"grad_norm": 4.263644218444824,
"learning_rate": 3.1354216867469884e-06,
"loss": 0.1364,
"step": 86000
},
{
"epoch": 1.194296,
"grad_norm": 3.34621524810791,
"learning_rate": 3.1273895582329323e-06,
"loss": 0.1215,
"step": 86100
},
{
"epoch": 1.195096,
"grad_norm": 5.672510623931885,
"learning_rate": 3.1193574297188754e-06,
"loss": 0.1323,
"step": 86200
},
{
"epoch": 1.195896,
"grad_norm": 7.123562812805176,
"learning_rate": 3.1113253012048193e-06,
"loss": 0.128,
"step": 86300
},
{
"epoch": 1.196696,
"grad_norm": 5.052427768707275,
"learning_rate": 3.1033734939759036e-06,
"loss": 0.1343,
"step": 86400
},
{
"epoch": 1.1974960000000001,
"grad_norm": 9.78446102142334,
"learning_rate": 3.0953413654618475e-06,
"loss": 0.1354,
"step": 86500
},
{
"epoch": 1.198296,
"grad_norm": 3.7782175540924072,
"learning_rate": 3.0873092369477914e-06,
"loss": 0.1369,
"step": 86600
},
{
"epoch": 1.199096,
"grad_norm": 3.037858247756958,
"learning_rate": 3.0792771084337354e-06,
"loss": 0.1368,
"step": 86700
},
{
"epoch": 1.199896,
"grad_norm": 6.603598117828369,
"learning_rate": 3.0712449799196793e-06,
"loss": 0.1313,
"step": 86800
},
{
"epoch": 1.200696,
"grad_norm": 5.845118522644043,
"learning_rate": 3.0632128514056224e-06,
"loss": 0.1381,
"step": 86900
},
{
"epoch": 1.2014960000000001,
"grad_norm": 5.331429958343506,
"learning_rate": 3.0551807228915663e-06,
"loss": 0.1394,
"step": 87000
},
{
"epoch": 1.202296,
"grad_norm": 2.769038677215576,
"learning_rate": 3.04714859437751e-06,
"loss": 0.138,
"step": 87100
},
{
"epoch": 1.203096,
"grad_norm": 4.4109320640563965,
"learning_rate": 3.039116465863454e-06,
"loss": 0.1355,
"step": 87200
},
{
"epoch": 1.203896,
"grad_norm": 5.107141494750977,
"learning_rate": 3.031084337349398e-06,
"loss": 0.1242,
"step": 87300
},
{
"epoch": 1.204696,
"grad_norm": 3.513662815093994,
"learning_rate": 3.0230522088353415e-06,
"loss": 0.1341,
"step": 87400
},
{
"epoch": 1.2054960000000001,
"grad_norm": 5.542943000793457,
"learning_rate": 3.0150200803212855e-06,
"loss": 0.1582,
"step": 87500
},
{
"epoch": 1.206296,
"grad_norm": 6.642317771911621,
"learning_rate": 3.0069879518072294e-06,
"loss": 0.1328,
"step": 87600
},
{
"epoch": 1.207096,
"grad_norm": 4.840209007263184,
"learning_rate": 2.9989558232931733e-06,
"loss": 0.1442,
"step": 87700
},
{
"epoch": 1.207896,
"grad_norm": 8.443116188049316,
"learning_rate": 2.9909236947791164e-06,
"loss": 0.1301,
"step": 87800
},
{
"epoch": 1.208696,
"grad_norm": 5.006475925445557,
"learning_rate": 2.9828915662650603e-06,
"loss": 0.137,
"step": 87900
},
{
"epoch": 1.209496,
"grad_norm": 16.58333969116211,
"learning_rate": 2.974859437751004e-06,
"loss": 0.141,
"step": 88000
},
{
"epoch": 1.209496,
"eval_test1_cer": 0.05272518485323773,
"eval_test1_cer_norm": 0.035959372823972914,
"eval_test1_loss": 0.1771300584077835,
"eval_test1_runtime": 3405.3955,
"eval_test1_samples_per_second": 0.734,
"eval_test1_steps_per_second": 0.184,
"eval_test1_wer": 0.15084987900521882,
"eval_test1_wer_norm": 0.0903883807241591,
"step": 88000
},
{
"epoch": 1.209496,
"eval_test2_cer": 0.10232108858774779,
"eval_test2_cer_norm": 0.08082681282925318,
"eval_test2_loss": 0.31750166416168213,
"eval_test2_runtime": 3462.1713,
"eval_test2_samples_per_second": 0.722,
"eval_test2_steps_per_second": 0.181,
"eval_test2_wer": 0.2193579766536965,
"eval_test2_wer_norm": 0.15279051111620445,
"step": 88000
},
{
"epoch": 1.210296,
"grad_norm": 7.229742527008057,
"learning_rate": 2.966827309236948e-06,
"loss": 0.1565,
"step": 88100
},
{
"epoch": 1.211096,
"grad_norm": 4.775784015655518,
"learning_rate": 2.958795180722892e-06,
"loss": 0.1191,
"step": 88200
},
{
"epoch": 1.211896,
"grad_norm": 5.914107799530029,
"learning_rate": 2.9507630522088355e-06,
"loss": 0.1272,
"step": 88300
},
{
"epoch": 1.212696,
"grad_norm": 3.42290997505188,
"learning_rate": 2.9427309236947795e-06,
"loss": 0.1533,
"step": 88400
},
{
"epoch": 1.213496,
"grad_norm": 5.825013637542725,
"learning_rate": 2.9347791164658634e-06,
"loss": 0.1337,
"step": 88500
},
{
"epoch": 1.214296,
"grad_norm": 2.929975748062134,
"learning_rate": 2.9267469879518073e-06,
"loss": 0.1363,
"step": 88600
},
{
"epoch": 1.215096,
"grad_norm": 4.544336318969727,
"learning_rate": 2.918714859437751e-06,
"loss": 0.1416,
"step": 88700
},
{
"epoch": 1.215896,
"grad_norm": 6.057349681854248,
"learning_rate": 2.910682730923695e-06,
"loss": 0.1298,
"step": 88800
},
{
"epoch": 1.216696,
"grad_norm": 4.51986026763916,
"learning_rate": 2.902650602409639e-06,
"loss": 0.1333,
"step": 88900
},
{
"epoch": 1.217496,
"grad_norm": 4.948803424835205,
"learning_rate": 2.8946184738955825e-06,
"loss": 0.134,
"step": 89000
},
{
"epoch": 1.218296,
"grad_norm": 4.863702774047852,
"learning_rate": 2.8865863453815264e-06,
"loss": 0.1378,
"step": 89100
},
{
"epoch": 1.219096,
"grad_norm": 4.049374580383301,
"learning_rate": 2.8785542168674704e-06,
"loss": 0.125,
"step": 89200
},
{
"epoch": 1.219896,
"grad_norm": 3.5720324516296387,
"learning_rate": 2.870522088353414e-06,
"loss": 0.1395,
"step": 89300
},
{
"epoch": 1.220696,
"grad_norm": 1.931492567062378,
"learning_rate": 2.8624899598393574e-06,
"loss": 0.131,
"step": 89400
},
{
"epoch": 1.221496,
"grad_norm": 6.574014663696289,
"learning_rate": 2.8544578313253013e-06,
"loss": 0.132,
"step": 89500
},
{
"epoch": 1.222296,
"grad_norm": 7.461375713348389,
"learning_rate": 2.846425702811245e-06,
"loss": 0.1289,
"step": 89600
},
{
"epoch": 1.223096,
"grad_norm": 5.367936611175537,
"learning_rate": 2.838393574297189e-06,
"loss": 0.1261,
"step": 89700
},
{
"epoch": 1.223896,
"grad_norm": 5.853065013885498,
"learning_rate": 2.830361445783133e-06,
"loss": 0.1161,
"step": 89800
},
{
"epoch": 1.224696,
"grad_norm": 4.45815896987915,
"learning_rate": 2.8223293172690765e-06,
"loss": 0.1285,
"step": 89900
},
{
"epoch": 1.225496,
"grad_norm": 5.519855499267578,
"learning_rate": 2.81429718875502e-06,
"loss": 0.1464,
"step": 90000
},
{
"epoch": 1.226296,
"grad_norm": 3.830179214477539,
"learning_rate": 2.806265060240964e-06,
"loss": 0.1278,
"step": 90100
},
{
"epoch": 1.227096,
"grad_norm": 3.485785722732544,
"learning_rate": 2.798232931726908e-06,
"loss": 0.1599,
"step": 90200
},
{
"epoch": 1.2278959999999999,
"grad_norm": 5.015017509460449,
"learning_rate": 2.7902008032128514e-06,
"loss": 0.1213,
"step": 90300
},
{
"epoch": 1.228696,
"grad_norm": 5.003856658935547,
"learning_rate": 2.7821686746987953e-06,
"loss": 0.1268,
"step": 90400
},
{
"epoch": 1.229496,
"grad_norm": 8.828622817993164,
"learning_rate": 2.77421686746988e-06,
"loss": 0.128,
"step": 90500
},
{
"epoch": 1.230296,
"grad_norm": 5.4841790199279785,
"learning_rate": 2.766184738955823e-06,
"loss": 0.1267,
"step": 90600
},
{
"epoch": 1.231096,
"grad_norm": 1.1707990169525146,
"learning_rate": 2.758152610441767e-06,
"loss": 0.1173,
"step": 90700
},
{
"epoch": 1.2318959999999999,
"grad_norm": 4.860354423522949,
"learning_rate": 2.750120481927711e-06,
"loss": 0.1248,
"step": 90800
},
{
"epoch": 1.232696,
"grad_norm": 4.709814071655273,
"learning_rate": 2.742088353413655e-06,
"loss": 0.1288,
"step": 90900
},
{
"epoch": 1.233496,
"grad_norm": 5.921780109405518,
"learning_rate": 2.7340562248995988e-06,
"loss": 0.1165,
"step": 91000
},
{
"epoch": 1.234296,
"grad_norm": 31.39832305908203,
"learning_rate": 2.7260240963855423e-06,
"loss": 0.1362,
"step": 91100
},
{
"epoch": 1.235096,
"grad_norm": 9.929832458496094,
"learning_rate": 2.717991967871486e-06,
"loss": 0.1356,
"step": 91200
},
{
"epoch": 1.2358959999999999,
"grad_norm": 6.635560035705566,
"learning_rate": 2.70995983935743e-06,
"loss": 0.1242,
"step": 91300
},
{
"epoch": 1.236696,
"grad_norm": 3.8171298503875732,
"learning_rate": 2.701927710843374e-06,
"loss": 0.1322,
"step": 91400
},
{
"epoch": 1.237496,
"grad_norm": 5.8909430503845215,
"learning_rate": 2.693895582329317e-06,
"loss": 0.1404,
"step": 91500
},
{
"epoch": 1.238296,
"grad_norm": 12.169204711914062,
"learning_rate": 2.685863453815261e-06,
"loss": 0.1312,
"step": 91600
},
{
"epoch": 1.239096,
"grad_norm": 2.663769483566284,
"learning_rate": 2.677831325301205e-06,
"loss": 0.1266,
"step": 91700
},
{
"epoch": 1.2398959999999999,
"grad_norm": 10.018877983093262,
"learning_rate": 2.669799196787149e-06,
"loss": 0.1279,
"step": 91800
},
{
"epoch": 1.240696,
"grad_norm": 3.6418004035949707,
"learning_rate": 2.661767068273093e-06,
"loss": 0.1233,
"step": 91900
},
{
"epoch": 1.241496,
"grad_norm": 6.836270332336426,
"learning_rate": 2.6537349397590363e-06,
"loss": 0.137,
"step": 92000
},
{
"epoch": 1.241496,
"eval_test1_cer": 0.04675013070430951,
"eval_test1_cer_norm": 0.031027444954018296,
"eval_test1_loss": 0.17643441259860992,
"eval_test1_runtime": 3383.8187,
"eval_test1_samples_per_second": 0.739,
"eval_test1_steps_per_second": 0.185,
"eval_test1_wer": 0.1424531327443949,
"eval_test1_wer_norm": 0.08281948625032877,
"step": 92000
},
{
"epoch": 1.241496,
"eval_test2_cer": 0.10391234554074738,
"eval_test2_cer_norm": 0.07987294700960644,
"eval_test2_loss": 0.31574827432632446,
"eval_test2_runtime": 3451.2496,
"eval_test2_samples_per_second": 0.724,
"eval_test2_steps_per_second": 0.181,
"eval_test2_wer": 0.22485122453650722,
"eval_test2_wer_norm": 0.1593513637405455,
"step": 92000
},
{
"epoch": 1.242296,
"grad_norm": 6.066310405731201,
"learning_rate": 2.64570281124498e-06,
"loss": 0.1255,
"step": 92100
},
{
"epoch": 1.243096,
"grad_norm": 4.294830799102783,
"learning_rate": 2.637670682730924e-06,
"loss": 0.1399,
"step": 92200
},
{
"epoch": 1.243896,
"grad_norm": 6.00723934173584,
"learning_rate": 2.629638554216868e-06,
"loss": 0.1446,
"step": 92300
},
{
"epoch": 1.244696,
"grad_norm": 10.248518943786621,
"learning_rate": 2.621606425702811e-06,
"loss": 0.1317,
"step": 92400
},
{
"epoch": 1.245496,
"grad_norm": 2.9732255935668945,
"learning_rate": 2.613654618473896e-06,
"loss": 0.1329,
"step": 92500
},
{
"epoch": 1.246296,
"grad_norm": 5.735791206359863,
"learning_rate": 2.6056224899598398e-06,
"loss": 0.1246,
"step": 92600
},
{
"epoch": 1.247096,
"grad_norm": 4.58491849899292,
"learning_rate": 2.5975903614457833e-06,
"loss": 0.118,
"step": 92700
},
{
"epoch": 1.247896,
"grad_norm": 6.321929931640625,
"learning_rate": 2.589558232931727e-06,
"loss": 0.1317,
"step": 92800
},
{
"epoch": 1.248696,
"grad_norm": 2.8514037132263184,
"learning_rate": 2.581526104417671e-06,
"loss": 0.124,
"step": 92900
},
{
"epoch": 1.249496,
"grad_norm": 6.127628803253174,
"learning_rate": 2.573493975903615e-06,
"loss": 0.1429,
"step": 93000
},
{
"epoch": 1.250296,
"grad_norm": 5.147157192230225,
"learning_rate": 2.565461847389558e-06,
"loss": 0.1294,
"step": 93100
},
{
"epoch": 1.251096,
"grad_norm": 1.7546809911727905,
"learning_rate": 2.557429718875502e-06,
"loss": 0.1228,
"step": 93200
},
{
"epoch": 1.251896,
"grad_norm": 5.040311813354492,
"learning_rate": 2.549397590361446e-06,
"loss": 0.1272,
"step": 93300
},
{
"epoch": 1.252696,
"grad_norm": 6.703260898590088,
"learning_rate": 2.54136546184739e-06,
"loss": 0.1312,
"step": 93400
},
{
"epoch": 1.253496,
"grad_norm": 4.4106974601745605,
"learning_rate": 2.5333333333333338e-06,
"loss": 0.1209,
"step": 93500
},
{
"epoch": 1.254296,
"grad_norm": 4.892759799957275,
"learning_rate": 2.5253012048192773e-06,
"loss": 0.1273,
"step": 93600
},
{
"epoch": 1.255096,
"grad_norm": 3.4537353515625,
"learning_rate": 2.517269076305221e-06,
"loss": 0.1292,
"step": 93700
},
{
"epoch": 1.255896,
"grad_norm": 6.59584379196167,
"learning_rate": 2.5092369477911647e-06,
"loss": 0.1194,
"step": 93800
},
{
"epoch": 1.256696,
"grad_norm": 7.888189792633057,
"learning_rate": 2.5012048192771086e-06,
"loss": 0.1292,
"step": 93900
},
{
"epoch": 1.257496,
"grad_norm": 3.4533464908599854,
"learning_rate": 2.4931726907630525e-06,
"loss": 0.1271,
"step": 94000
},
{
"epoch": 1.258296,
"grad_norm": 1.2868270874023438,
"learning_rate": 2.4851405622489965e-06,
"loss": 0.1416,
"step": 94100
},
{
"epoch": 1.259096,
"grad_norm": 4.686305999755859,
"learning_rate": 2.47710843373494e-06,
"loss": 0.135,
"step": 94200
},
{
"epoch": 1.259896,
"grad_norm": 2.333994150161743,
"learning_rate": 2.469076305220884e-06,
"loss": 0.1143,
"step": 94300
},
{
"epoch": 1.260696,
"grad_norm": 24.702146530151367,
"learning_rate": 2.4610441767068274e-06,
"loss": 0.122,
"step": 94400
},
{
"epoch": 1.261496,
"grad_norm": 2.577688694000244,
"learning_rate": 2.4530120481927713e-06,
"loss": 0.1323,
"step": 94500
},
{
"epoch": 1.262296,
"grad_norm": 4.011758804321289,
"learning_rate": 2.4449799196787148e-06,
"loss": 0.1375,
"step": 94600
},
{
"epoch": 1.263096,
"grad_norm": 3.8181052207946777,
"learning_rate": 2.4369477911646587e-06,
"loss": 0.1364,
"step": 94700
},
{
"epoch": 1.263896,
"grad_norm": 5.570916652679443,
"learning_rate": 2.4289959839357434e-06,
"loss": 0.1334,
"step": 94800
},
{
"epoch": 1.264696,
"grad_norm": 5.255520820617676,
"learning_rate": 2.420963855421687e-06,
"loss": 0.1346,
"step": 94900
},
{
"epoch": 1.265496,
"grad_norm": 6.02402400970459,
"learning_rate": 2.412931726907631e-06,
"loss": 0.1299,
"step": 95000
},
{
"epoch": 1.266296,
"grad_norm": 5.61665153503418,
"learning_rate": 2.4048995983935744e-06,
"loss": 0.1332,
"step": 95100
},
{
"epoch": 1.267096,
"grad_norm": 3.135876417160034,
"learning_rate": 2.3968674698795183e-06,
"loss": 0.1466,
"step": 95200
},
{
"epoch": 1.267896,
"grad_norm": 3.9527804851531982,
"learning_rate": 2.3888353413654618e-06,
"loss": 0.118,
"step": 95300
},
{
"epoch": 1.268696,
"grad_norm": 8.5547513961792,
"learning_rate": 2.3808032128514057e-06,
"loss": 0.1289,
"step": 95400
},
{
"epoch": 1.269496,
"grad_norm": 4.350017070770264,
"learning_rate": 2.3727710843373496e-06,
"loss": 0.1485,
"step": 95500
},
{
"epoch": 1.270296,
"grad_norm": 7.225109100341797,
"learning_rate": 2.3647389558232935e-06,
"loss": 0.1724,
"step": 95600
},
{
"epoch": 1.271096,
"grad_norm": 3.152327060699463,
"learning_rate": 2.356706827309237e-06,
"loss": 0.151,
"step": 95700
},
{
"epoch": 1.271896,
"grad_norm": 9.343049049377441,
"learning_rate": 2.348674698795181e-06,
"loss": 0.1219,
"step": 95800
},
{
"epoch": 1.272696,
"grad_norm": 3.7528295516967773,
"learning_rate": 2.340642570281125e-06,
"loss": 0.1011,
"step": 95900
},
{
"epoch": 1.273496,
"grad_norm": 1.7038061618804932,
"learning_rate": 2.3326104417670684e-06,
"loss": 0.1287,
"step": 96000
},
{
"epoch": 1.273496,
"eval_test1_cer": 0.04843528269474942,
"eval_test1_cer_norm": 0.03259298388839532,
"eval_test1_loss": 0.17615145444869995,
"eval_test1_runtime": 3390.194,
"eval_test1_samples_per_second": 0.737,
"eval_test1_steps_per_second": 0.184,
"eval_test1_wer": 0.14195749147205458,
"eval_test1_wer_norm": 0.08346240392764254,
"step": 96000
},
{
"epoch": 1.273496,
"eval_test2_cer": 0.10040784709000634,
"eval_test2_cer_norm": 0.07324914781530834,
"eval_test2_loss": 0.31477001309394836,
"eval_test2_runtime": 3429.6171,
"eval_test2_samples_per_second": 0.729,
"eval_test2_steps_per_second": 0.182,
"eval_test2_wer": 0.21283474479285877,
"eval_test2_wer_norm": 0.14789135915654367,
"step": 96000
},
{
"epoch": 1.274296,
"grad_norm": 4.975541591644287,
"learning_rate": 2.3245783132530123e-06,
"loss": 0.1545,
"step": 96100
},
{
"epoch": 1.275096,
"grad_norm": 2.7354822158813477,
"learning_rate": 2.316546184738956e-06,
"loss": 0.1379,
"step": 96200
},
{
"epoch": 1.275896,
"grad_norm": 4.287178993225098,
"learning_rate": 2.3085140562248997e-06,
"loss": 0.1109,
"step": 96300
},
{
"epoch": 1.276696,
"grad_norm": 7.494585990905762,
"learning_rate": 2.3004819277108436e-06,
"loss": 0.1361,
"step": 96400
},
{
"epoch": 1.277496,
"grad_norm": 5.967809677124023,
"learning_rate": 2.292449799196787e-06,
"loss": 0.1282,
"step": 96500
},
{
"epoch": 1.278296,
"grad_norm": 6.256405830383301,
"learning_rate": 2.284417670682731e-06,
"loss": 0.1254,
"step": 96600
},
{
"epoch": 1.279096,
"grad_norm": 4.7275214195251465,
"learning_rate": 2.276385542168675e-06,
"loss": 0.1125,
"step": 96700
},
{
"epoch": 1.279896,
"grad_norm": 6.460155010223389,
"learning_rate": 2.2685140562248997e-06,
"loss": 0.1531,
"step": 96800
},
{
"epoch": 1.280696,
"grad_norm": 7.814698696136475,
"learning_rate": 2.2604819277108436e-06,
"loss": 0.1353,
"step": 96900
},
{
"epoch": 1.281496,
"grad_norm": 2.7123336791992188,
"learning_rate": 2.252449799196787e-06,
"loss": 0.1223,
"step": 97000
},
{
"epoch": 1.282296,
"grad_norm": 4.504587650299072,
"learning_rate": 2.244417670682731e-06,
"loss": 0.1358,
"step": 97100
},
{
"epoch": 1.283096,
"grad_norm": 7.585097789764404,
"learning_rate": 2.236385542168675e-06,
"loss": 0.1378,
"step": 97200
},
{
"epoch": 1.283896,
"grad_norm": 3.129997491836548,
"learning_rate": 2.228353413654619e-06,
"loss": 0.1234,
"step": 97300
},
{
"epoch": 1.284696,
"grad_norm": 6.103450775146484,
"learning_rate": 2.2203212851405628e-06,
"loss": 0.1188,
"step": 97400
},
{
"epoch": 1.285496,
"grad_norm": 5.75684928894043,
"learning_rate": 2.2122891566265063e-06,
"loss": 0.118,
"step": 97500
},
{
"epoch": 1.286296,
"grad_norm": 4.268872261047363,
"learning_rate": 2.20425702811245e-06,
"loss": 0.1242,
"step": 97600
},
{
"epoch": 1.287096,
"grad_norm": 10.057229042053223,
"learning_rate": 2.1962248995983937e-06,
"loss": 0.107,
"step": 97700
},
{
"epoch": 1.287896,
"grad_norm": 6.790497303009033,
"learning_rate": 2.1881927710843376e-06,
"loss": 0.1169,
"step": 97800
},
{
"epoch": 1.288696,
"grad_norm": 7.399913311004639,
"learning_rate": 2.180160642570281e-06,
"loss": 0.1389,
"step": 97900
},
{
"epoch": 1.289496,
"grad_norm": 8.95304012298584,
"learning_rate": 2.172128514056225e-06,
"loss": 0.1119,
"step": 98000
},
{
"epoch": 1.290296,
"grad_norm": 3.782831907272339,
"learning_rate": 2.164096385542169e-06,
"loss": 0.1158,
"step": 98100
},
{
"epoch": 1.291096,
"grad_norm": 9.19373893737793,
"learning_rate": 2.1560642570281124e-06,
"loss": 0.1373,
"step": 98200
},
{
"epoch": 1.291896,
"grad_norm": 8.121448516845703,
"learning_rate": 2.1480321285140563e-06,
"loss": 0.1379,
"step": 98300
},
{
"epoch": 1.292696,
"grad_norm": 5.722080707550049,
"learning_rate": 2.1400000000000003e-06,
"loss": 0.1171,
"step": 98400
},
{
"epoch": 1.293496,
"grad_norm": 4.1129255294799805,
"learning_rate": 2.131967871485944e-06,
"loss": 0.1302,
"step": 98500
},
{
"epoch": 1.2942960000000001,
"grad_norm": 3.102550983428955,
"learning_rate": 2.1239357429718877e-06,
"loss": 0.1256,
"step": 98600
},
{
"epoch": 1.295096,
"grad_norm": 5.705405235290527,
"learning_rate": 2.1159036144578316e-06,
"loss": 0.1103,
"step": 98700
},
{
"epoch": 1.295896,
"grad_norm": 5.043685436248779,
"learning_rate": 2.107871485943775e-06,
"loss": 0.1207,
"step": 98800
},
{
"epoch": 1.296696,
"grad_norm": 5.753846645355225,
"learning_rate": 2.099839357429719e-06,
"loss": 0.1195,
"step": 98900
},
{
"epoch": 1.297496,
"grad_norm": 5.277960300445557,
"learning_rate": 2.091807228915663e-06,
"loss": 0.1326,
"step": 99000
},
{
"epoch": 1.2982960000000001,
"grad_norm": 4.641193866729736,
"learning_rate": 2.0837751004016064e-06,
"loss": 0.1409,
"step": 99100
},
{
"epoch": 1.299096,
"grad_norm": 6.307822227478027,
"learning_rate": 2.0757429718875504e-06,
"loss": 0.1271,
"step": 99200
},
{
"epoch": 1.299896,
"grad_norm": 4.50252628326416,
"learning_rate": 2.0677108433734943e-06,
"loss": 0.1242,
"step": 99300
},
{
"epoch": 1.300696,
"grad_norm": 6.755404949188232,
"learning_rate": 2.059678714859438e-06,
"loss": 0.1349,
"step": 99400
},
{
"epoch": 1.301496,
"grad_norm": 4.805443286895752,
"learning_rate": 2.0516465863453817e-06,
"loss": 0.1227,
"step": 99500
},
{
"epoch": 1.3022960000000001,
"grad_norm": 3.4150850772857666,
"learning_rate": 2.0436144578313256e-06,
"loss": 0.1147,
"step": 99600
},
{
"epoch": 1.303096,
"grad_norm": 6.213440895080566,
"learning_rate": 2.035582329317269e-06,
"loss": 0.1236,
"step": 99700
},
{
"epoch": 1.303896,
"grad_norm": 6.957757472991943,
"learning_rate": 2.027550200803213e-06,
"loss": 0.1323,
"step": 99800
},
{
"epoch": 1.304696,
"grad_norm": 7.19875955581665,
"learning_rate": 2.019518072289157e-06,
"loss": 0.1283,
"step": 99900
},
{
"epoch": 1.305496,
"grad_norm": 2.5074002742767334,
"learning_rate": 2.0114859437751004e-06,
"loss": 0.1128,
"step": 100000
},
{
"epoch": 1.305496,
"eval_test1_cer": 0.0477444170587796,
"eval_test1_cer_norm": 0.03308281508872188,
"eval_test1_loss": 0.17668992280960083,
"eval_test1_runtime": 3382.0305,
"eval_test1_samples_per_second": 0.739,
"eval_test1_steps_per_second": 0.185,
"eval_test1_wer": 0.1387795562552844,
"eval_test1_wer_norm": 0.07951722727139893,
"step": 100000
},
{
"epoch": 1.305496,
"eval_test2_cer": 0.10429499384029567,
"eval_test2_cer_norm": 0.0824295010845987,
"eval_test2_loss": 0.3178161084651947,
"eval_test2_runtime": 3460.99,
"eval_test2_samples_per_second": 0.722,
"eval_test2_steps_per_second": 0.181,
"eval_test2_wer": 0.2208743419546807,
"eval_test2_wer_norm": 0.15585606234242494,
"step": 100000
},
{
"epoch": 1.3062960000000001,
"grad_norm": 4.352635383605957,
"learning_rate": 2.0034538152610444e-06,
"loss": 0.1269,
"step": 100100
},
{
"epoch": 1.307096,
"grad_norm": 6.455359935760498,
"learning_rate": 1.995421686746988e-06,
"loss": 0.1228,
"step": 100200
},
{
"epoch": 1.307896,
"grad_norm": 4.56587028503418,
"learning_rate": 1.9873895582329318e-06,
"loss": 0.1302,
"step": 100300
},
{
"epoch": 1.308696,
"grad_norm": 6.275040149688721,
"learning_rate": 1.9793574297188757e-06,
"loss": 0.1316,
"step": 100400
},
{
"epoch": 1.309496,
"grad_norm": 2.4755728244781494,
"learning_rate": 1.9713253012048196e-06,
"loss": 0.1324,
"step": 100500
},
{
"epoch": 1.3102960000000001,
"grad_norm": 5.687671184539795,
"learning_rate": 1.963293172690763e-06,
"loss": 0.1296,
"step": 100600
},
{
"epoch": 1.311096,
"grad_norm": 3.915098190307617,
"learning_rate": 1.955261044176707e-06,
"loss": 0.1198,
"step": 100700
},
{
"epoch": 1.311896,
"grad_norm": 5.507267475128174,
"learning_rate": 1.947228915662651e-06,
"loss": 0.1308,
"step": 100800
},
{
"epoch": 1.3126959999999999,
"grad_norm": 6.799343109130859,
"learning_rate": 1.939277108433735e-06,
"loss": 0.1106,
"step": 100900
},
{
"epoch": 1.313496,
"grad_norm": 2.4747636318206787,
"learning_rate": 1.9312449799196788e-06,
"loss": 0.1198,
"step": 101000
},
{
"epoch": 1.3142960000000001,
"grad_norm": 1.4364855289459229,
"learning_rate": 1.9232128514056227e-06,
"loss": 0.1251,
"step": 101100
},
{
"epoch": 1.315096,
"grad_norm": 3.4942729473114014,
"learning_rate": 1.9151807228915666e-06,
"loss": 0.105,
"step": 101200
},
{
"epoch": 1.315896,
"grad_norm": 7.960775852203369,
"learning_rate": 1.9071485943775103e-06,
"loss": 0.121,
"step": 101300
},
{
"epoch": 1.3166959999999999,
"grad_norm": 4.901790618896484,
"learning_rate": 1.899116465863454e-06,
"loss": 0.1248,
"step": 101400
},
{
"epoch": 1.317496,
"grad_norm": 3.4517650604248047,
"learning_rate": 1.891084337349398e-06,
"loss": 0.1101,
"step": 101500
},
{
"epoch": 1.3182960000000001,
"grad_norm": 2.9589340686798096,
"learning_rate": 1.8830522088353414e-06,
"loss": 0.1569,
"step": 101600
},
{
"epoch": 1.319096,
"grad_norm": 4.821226119995117,
"learning_rate": 1.8750200803212854e-06,
"loss": 0.1248,
"step": 101700
},
{
"epoch": 1.319896,
"grad_norm": 5.956510543823242,
"learning_rate": 1.866987951807229e-06,
"loss": 0.129,
"step": 101800
},
{
"epoch": 1.3206959999999999,
"grad_norm": 8.236087799072266,
"learning_rate": 1.8589558232931728e-06,
"loss": 0.1288,
"step": 101900
},
{
"epoch": 1.321496,
"grad_norm": 7.885124206542969,
"learning_rate": 1.8509236947791165e-06,
"loss": 0.133,
"step": 102000
},
{
"epoch": 1.3222960000000001,
"grad_norm": 4.618575096130371,
"learning_rate": 1.8428915662650604e-06,
"loss": 0.131,
"step": 102100
},
{
"epoch": 1.323096,
"grad_norm": 3.3860082626342773,
"learning_rate": 1.8348594377510043e-06,
"loss": 0.1275,
"step": 102200
},
{
"epoch": 1.323896,
"grad_norm": 0.5851134061813354,
"learning_rate": 1.8268273092369478e-06,
"loss": 0.1163,
"step": 102300
},
{
"epoch": 1.3246959999999999,
"grad_norm": 8.403120040893555,
"learning_rate": 1.8187951807228917e-06,
"loss": 0.142,
"step": 102400
},
{
"epoch": 1.325496,
"grad_norm": 4.019391059875488,
"learning_rate": 1.8107630522088354e-06,
"loss": 0.1234,
"step": 102500
},
{
"epoch": 1.326296,
"grad_norm": 7.64513635635376,
"learning_rate": 1.8028112449799198e-06,
"loss": 0.1202,
"step": 102600
},
{
"epoch": 1.327096,
"grad_norm": 4.8131489753723145,
"learning_rate": 1.7947791164658637e-06,
"loss": 0.1169,
"step": 102700
},
{
"epoch": 1.327896,
"grad_norm": 7.043356895446777,
"learning_rate": 1.7867469879518074e-06,
"loss": 0.1209,
"step": 102800
},
{
"epoch": 1.3286959999999999,
"grad_norm": 4.337855815887451,
"learning_rate": 1.7787148594377513e-06,
"loss": 0.1365,
"step": 102900
},
{
"epoch": 1.329496,
"grad_norm": 6.657166481018066,
"learning_rate": 1.7706827309236948e-06,
"loss": 0.1256,
"step": 103000
},
{
"epoch": 1.330296,
"grad_norm": 4.905508041381836,
"learning_rate": 1.7626506024096387e-06,
"loss": 0.1118,
"step": 103100
},
{
"epoch": 1.331096,
"grad_norm": 5.122402667999268,
"learning_rate": 1.7546184738955824e-06,
"loss": 0.1276,
"step": 103200
},
{
"epoch": 1.331896,
"grad_norm": 4.78582239151001,
"learning_rate": 1.7465863453815264e-06,
"loss": 0.1194,
"step": 103300
},
{
"epoch": 1.3326959999999999,
"grad_norm": 5.943844318389893,
"learning_rate": 1.7385542168674698e-06,
"loss": 0.1206,
"step": 103400
},
{
"epoch": 1.333496,
"grad_norm": 8.206818580627441,
"learning_rate": 1.7305220883534138e-06,
"loss": 0.1151,
"step": 103500
},
{
"epoch": 1.334296,
"grad_norm": 5.556063175201416,
"learning_rate": 1.7224899598393577e-06,
"loss": 0.1114,
"step": 103600
},
{
"epoch": 1.335096,
"grad_norm": 7.587924957275391,
"learning_rate": 1.7144578313253014e-06,
"loss": 0.1207,
"step": 103700
},
{
"epoch": 1.335896,
"grad_norm": 7.417220592498779,
"learning_rate": 1.706425702811245e-06,
"loss": 0.1163,
"step": 103800
},
{
"epoch": 1.3366959999999999,
"grad_norm": 7.316643238067627,
"learning_rate": 1.6983935742971888e-06,
"loss": 0.1157,
"step": 103900
},
{
"epoch": 1.337496,
"grad_norm": 1.7829804420471191,
"learning_rate": 1.6903614457831327e-06,
"loss": 0.1155,
"step": 104000
},
{
"epoch": 1.337496,
"eval_test1_cer": 0.045685824184031665,
"eval_test1_cer_norm": 0.03030710495353807,
"eval_test1_loss": 0.17226466536521912,
"eval_test1_runtime": 3385.0143,
"eval_test1_samples_per_second": 0.739,
"eval_test1_steps_per_second": 0.185,
"eval_test1_wer": 0.13615557304877693,
"eval_test1_wer_norm": 0.07755925070867062,
"step": 104000
},
{
"epoch": 1.337496,
"eval_test2_cer": 0.10362769253742488,
"eval_test2_cer_norm": 0.08312674310505114,
"eval_test2_loss": 0.31396326422691345,
"eval_test2_runtime": 3448.444,
"eval_test2_samples_per_second": 0.725,
"eval_test2_steps_per_second": 0.181,
"eval_test2_wer": 0.21904325932707713,
"eval_test2_wer_norm": 0.15513981205592484,
"step": 104000
},
{
"epoch": 1.338296,
"grad_norm": 3.8256428241729736,
"learning_rate": 1.6823293172690762e-06,
"loss": 0.1251,
"step": 104100
},
{
"epoch": 1.339096,
"grad_norm": 4.764732837677002,
"learning_rate": 1.6742971887550201e-06,
"loss": 0.1183,
"step": 104200
},
{
"epoch": 1.339896,
"grad_norm": 3.5565683841705322,
"learning_rate": 1.666265060240964e-06,
"loss": 0.1152,
"step": 104300
},
{
"epoch": 1.3406959999999999,
"grad_norm": 5.896035671234131,
"learning_rate": 1.6582329317269078e-06,
"loss": 0.136,
"step": 104400
},
{
"epoch": 1.341496,
"grad_norm": 6.5424628257751465,
"learning_rate": 1.6502008032128517e-06,
"loss": 0.128,
"step": 104500
},
{
"epoch": 1.342296,
"grad_norm": 5.04475736618042,
"learning_rate": 1.6421686746987952e-06,
"loss": 0.1149,
"step": 104600
},
{
"epoch": 1.343096,
"grad_norm": 5.732762813568115,
"learning_rate": 1.6341365461847391e-06,
"loss": 0.131,
"step": 104700
},
{
"epoch": 1.343896,
"grad_norm": 5.360403060913086,
"learning_rate": 1.6261044176706828e-06,
"loss": 0.1283,
"step": 104800
},
{
"epoch": 1.344696,
"grad_norm": 6.674147129058838,
"learning_rate": 1.6180722891566267e-06,
"loss": 0.1357,
"step": 104900
},
{
"epoch": 1.345496,
"grad_norm": 3.1230456829071045,
"learning_rate": 1.6100401606425702e-06,
"loss": 0.1237,
"step": 105000
},
{
"epoch": 1.346296,
"grad_norm": 6.3545756340026855,
"learning_rate": 1.6020080321285142e-06,
"loss": 0.1248,
"step": 105100
},
{
"epoch": 1.347096,
"grad_norm": 9.923174858093262,
"learning_rate": 1.593975903614458e-06,
"loss": 0.1382,
"step": 105200
},
{
"epoch": 1.347896,
"grad_norm": 3.6945223808288574,
"learning_rate": 1.5859437751004018e-06,
"loss": 0.1244,
"step": 105300
},
{
"epoch": 1.348696,
"grad_norm": 8.398360252380371,
"learning_rate": 1.5779116465863457e-06,
"loss": 0.123,
"step": 105400
},
{
"epoch": 1.349496,
"grad_norm": 1.4128965139389038,
"learning_rate": 1.5698795180722892e-06,
"loss": 0.1175,
"step": 105500
},
{
"epoch": 1.350296,
"grad_norm": 11.629409790039062,
"learning_rate": 1.5618473895582331e-06,
"loss": 0.1117,
"step": 105600
},
{
"epoch": 1.351096,
"grad_norm": 4.094301700592041,
"learning_rate": 1.5538152610441768e-06,
"loss": 0.1193,
"step": 105700
},
{
"epoch": 1.351896,
"grad_norm": 4.269906044006348,
"learning_rate": 1.5457831325301205e-06,
"loss": 0.1256,
"step": 105800
},
{
"epoch": 1.352696,
"grad_norm": 5.27727746963501,
"learning_rate": 1.5377510040160642e-06,
"loss": 0.1224,
"step": 105900
},
{
"epoch": 1.353496,
"grad_norm": 4.353626728057861,
"learning_rate": 1.5297188755020082e-06,
"loss": 0.1196,
"step": 106000
},
{
"epoch": 1.354296,
"grad_norm": 3.226297616958618,
"learning_rate": 1.521686746987952e-06,
"loss": 0.1364,
"step": 106100
},
{
"epoch": 1.355096,
"grad_norm": 0.7694222331047058,
"learning_rate": 1.5136546184738956e-06,
"loss": 0.103,
"step": 106200
},
{
"epoch": 1.355896,
"grad_norm": 3.6190733909606934,
"learning_rate": 1.5056224899598395e-06,
"loss": 0.1235,
"step": 106300
},
{
"epoch": 1.356696,
"grad_norm": 3.509673833847046,
"learning_rate": 1.4975903614457832e-06,
"loss": 0.121,
"step": 106400
},
{
"epoch": 1.357496,
"grad_norm": 8.607564926147461,
"learning_rate": 1.4895582329317271e-06,
"loss": 0.1318,
"step": 106500
},
{
"epoch": 1.358296,
"grad_norm": 3.327017307281494,
"learning_rate": 1.4815261044176706e-06,
"loss": 0.1283,
"step": 106600
},
{
"epoch": 1.359096,
"grad_norm": 8.573492050170898,
"learning_rate": 1.4735742971887552e-06,
"loss": 0.1222,
"step": 106700
},
{
"epoch": 1.359896,
"grad_norm": 4.089103698730469,
"learning_rate": 1.465542168674699e-06,
"loss": 0.1257,
"step": 106800
},
{
"epoch": 1.360696,
"grad_norm": 3.942911386489868,
"learning_rate": 1.4575100401606426e-06,
"loss": 0.1205,
"step": 106900
},
{
"epoch": 1.361496,
"grad_norm": 6.010968208312988,
"learning_rate": 1.4494779116465865e-06,
"loss": 0.1295,
"step": 107000
},
{
"epoch": 1.362296,
"grad_norm": 6.7945756912231445,
"learning_rate": 1.4414457831325302e-06,
"loss": 0.1246,
"step": 107100
},
{
"epoch": 1.363096,
"grad_norm": 2.255643129348755,
"learning_rate": 1.4334136546184741e-06,
"loss": 0.1254,
"step": 107200
},
{
"epoch": 1.363896,
"grad_norm": 4.047929763793945,
"learning_rate": 1.4253815261044176e-06,
"loss": 0.1197,
"step": 107300
},
{
"epoch": 1.364696,
"grad_norm": 4.005350112915039,
"learning_rate": 1.4173493975903615e-06,
"loss": 0.1261,
"step": 107400
},
{
"epoch": 1.365496,
"grad_norm": 5.783274173736572,
"learning_rate": 1.4093172690763055e-06,
"loss": 0.1157,
"step": 107500
},
{
"epoch": 1.366296,
"grad_norm": 5.45566463470459,
"learning_rate": 1.4012851405622492e-06,
"loss": 0.125,
"step": 107600
},
{
"epoch": 1.367096,
"grad_norm": 4.851735591888428,
"learning_rate": 1.3932530120481929e-06,
"loss": 0.1242,
"step": 107700
},
{
"epoch": 1.367896,
"grad_norm": 5.266098976135254,
"learning_rate": 1.3852208835341366e-06,
"loss": 0.1267,
"step": 107800
},
{
"epoch": 1.368696,
"grad_norm": 4.1477742195129395,
"learning_rate": 1.3771887550200805e-06,
"loss": 0.1158,
"step": 107900
},
{
"epoch": 1.369496,
"grad_norm": 12.537534713745117,
"learning_rate": 1.369156626506024e-06,
"loss": 0.1114,
"step": 108000
},
{
"epoch": 1.369496,
"eval_test1_cer": 0.041395922025543354,
"eval_test1_cer_norm": 0.027075179484716785,
"eval_test1_loss": 0.1729966700077057,
"eval_test1_runtime": 3379.841,
"eval_test1_samples_per_second": 0.74,
"eval_test1_steps_per_second": 0.185,
"eval_test1_wer": 0.12840024490509927,
"eval_test1_wer_norm": 0.06993190917326632,
"step": 108000
},
{
"epoch": 1.369496,
"eval_test2_cer": 0.1033523724194572,
"eval_test2_cer_norm": 0.077180818097304,
"eval_test2_loss": 0.3105059862136841,
"eval_test2_runtime": 3449.5535,
"eval_test2_samples_per_second": 0.725,
"eval_test2_steps_per_second": 0.181,
"eval_test2_wer": 0.2153238727397574,
"eval_test2_wer_norm": 0.15038391015356406,
"step": 108000
},
{
"epoch": 1.370296,
"grad_norm": 2.571197986602783,
"learning_rate": 1.361124497991968e-06,
"loss": 0.1133,
"step": 108100
},
{
"epoch": 1.371096,
"grad_norm": 2.839953660964966,
"learning_rate": 1.3531726907630524e-06,
"loss": 0.1225,
"step": 108200
},
{
"epoch": 1.371896,
"grad_norm": 2.108365535736084,
"learning_rate": 1.345140562248996e-06,
"loss": 0.1111,
"step": 108300
},
{
"epoch": 1.372696,
"grad_norm": 4.538883686065674,
"learning_rate": 1.3371084337349399e-06,
"loss": 0.0988,
"step": 108400
},
{
"epoch": 1.373496,
"grad_norm": 5.8930277824401855,
"learning_rate": 1.3290763052208836e-06,
"loss": 0.1124,
"step": 108500
},
{
"epoch": 1.374296,
"grad_norm": 3.4291086196899414,
"learning_rate": 1.3210441767068275e-06,
"loss": 0.0998,
"step": 108600
},
{
"epoch": 1.375096,
"grad_norm": 3.634300947189331,
"learning_rate": 1.313012048192771e-06,
"loss": 0.123,
"step": 108700
},
{
"epoch": 1.375896,
"grad_norm": 10.641935348510742,
"learning_rate": 1.304979919678715e-06,
"loss": 0.1331,
"step": 108800
},
{
"epoch": 1.376696,
"grad_norm": 3.8561360836029053,
"learning_rate": 1.2969477911646588e-06,
"loss": 0.1386,
"step": 108900
},
{
"epoch": 1.377496,
"grad_norm": 5.578998565673828,
"learning_rate": 1.2889156626506025e-06,
"loss": 0.1252,
"step": 109000
},
{
"epoch": 1.378296,
"grad_norm": 6.097825050354004,
"learning_rate": 1.2808835341365464e-06,
"loss": 0.1226,
"step": 109100
},
{
"epoch": 1.379096,
"grad_norm": 7.930078983306885,
"learning_rate": 1.27285140562249e-06,
"loss": 0.1324,
"step": 109200
},
{
"epoch": 1.379896,
"grad_norm": 4.191389083862305,
"learning_rate": 1.2648192771084339e-06,
"loss": 0.1265,
"step": 109300
},
{
"epoch": 1.380696,
"grad_norm": 4.194836139678955,
"learning_rate": 1.2567871485943776e-06,
"loss": 0.1131,
"step": 109400
},
{
"epoch": 1.381496,
"grad_norm": 0.9715979695320129,
"learning_rate": 1.2487550200803215e-06,
"loss": 0.1149,
"step": 109500
},
{
"epoch": 1.382296,
"grad_norm": 2.0711517333984375,
"learning_rate": 1.2407228915662652e-06,
"loss": 0.1196,
"step": 109600
},
{
"epoch": 1.383096,
"grad_norm": 3.178311824798584,
"learning_rate": 1.232690763052209e-06,
"loss": 0.1116,
"step": 109700
},
{
"epoch": 1.383896,
"grad_norm": 4.997854232788086,
"learning_rate": 1.2246586345381526e-06,
"loss": 0.1215,
"step": 109800
},
{
"epoch": 1.384696,
"grad_norm": 4.431860446929932,
"learning_rate": 1.2166265060240963e-06,
"loss": 0.1129,
"step": 109900
},
{
"epoch": 1.385496,
"grad_norm": 4.320796966552734,
"learning_rate": 1.2085943775100402e-06,
"loss": 0.1375,
"step": 110000
},
{
"epoch": 1.386296,
"grad_norm": 7.0986552238464355,
"learning_rate": 1.2005622489959842e-06,
"loss": 0.1051,
"step": 110100
},
{
"epoch": 1.387096,
"grad_norm": 2.893299102783203,
"learning_rate": 1.1925301204819279e-06,
"loss": 0.1133,
"step": 110200
},
{
"epoch": 1.387896,
"grad_norm": 1.8434147834777832,
"learning_rate": 1.1844979919678716e-06,
"loss": 0.1099,
"step": 110300
},
{
"epoch": 1.388696,
"grad_norm": 3.1360116004943848,
"learning_rate": 1.1764658634538153e-06,
"loss": 0.1276,
"step": 110400
},
{
"epoch": 1.389496,
"grad_norm": 2.8671581745147705,
"learning_rate": 1.1684337349397592e-06,
"loss": 0.1113,
"step": 110500
},
{
"epoch": 1.390296,
"grad_norm": 8.354147911071777,
"learning_rate": 1.160401606425703e-06,
"loss": 0.1085,
"step": 110600
},
{
"epoch": 1.391096,
"grad_norm": 14.839973449707031,
"learning_rate": 1.1523694779116466e-06,
"loss": 0.1096,
"step": 110700
},
{
"epoch": 1.391896,
"grad_norm": 3.917928695678711,
"learning_rate": 1.1443373493975903e-06,
"loss": 0.1219,
"step": 110800
},
{
"epoch": 1.392696,
"grad_norm": 6.117668151855469,
"learning_rate": 1.1363052208835343e-06,
"loss": 0.1115,
"step": 110900
},
{
"epoch": 1.393496,
"grad_norm": 1.9348158836364746,
"learning_rate": 1.128273092369478e-06,
"loss": 0.1219,
"step": 111000
},
{
"epoch": 1.394296,
"grad_norm": 5.231213569641113,
"learning_rate": 1.1202409638554219e-06,
"loss": 0.1178,
"step": 111100
},
{
"epoch": 1.3950960000000001,
"grad_norm": 5.860185623168945,
"learning_rate": 1.1122088353413656e-06,
"loss": 0.1136,
"step": 111200
},
{
"epoch": 1.395896,
"grad_norm": 4.386104106903076,
"learning_rate": 1.10425702811245e-06,
"loss": 0.1159,
"step": 111300
},
{
"epoch": 1.396696,
"grad_norm": 1.6194918155670166,
"learning_rate": 1.0962248995983938e-06,
"loss": 0.1187,
"step": 111400
},
{
"epoch": 1.397496,
"grad_norm": 5.940235137939453,
"learning_rate": 1.0881927710843375e-06,
"loss": 0.1157,
"step": 111500
},
{
"epoch": 1.398296,
"grad_norm": 4.942705154418945,
"learning_rate": 1.0801606425702812e-06,
"loss": 0.1005,
"step": 111600
},
{
"epoch": 1.3990960000000001,
"grad_norm": 5.09475564956665,
"learning_rate": 1.072128514056225e-06,
"loss": 0.1204,
"step": 111700
},
{
"epoch": 1.399896,
"grad_norm": 5.27365255355835,
"learning_rate": 1.0640963855421687e-06,
"loss": 0.127,
"step": 111800
},
{
"epoch": 1.400696,
"grad_norm": 6.454215049743652,
"learning_rate": 1.0560642570281126e-06,
"loss": 0.1025,
"step": 111900
},
{
"epoch": 1.401496,
"grad_norm": 5.676671028137207,
"learning_rate": 1.0480321285140563e-06,
"loss": 0.1141,
"step": 112000
},
{
"epoch": 1.401496,
"eval_test1_cer": 0.04145660616924341,
"eval_test1_cer_norm": 0.02695992508463995,
"eval_test1_loss": 0.17103232443332672,
"eval_test1_runtime": 2490.4053,
"eval_test1_samples_per_second": 1.004,
"eval_test1_steps_per_second": 0.251,
"eval_test1_wer": 0.12912912912912913,
"eval_test1_wer_norm": 0.07016569741956223,
"step": 112000
},
{
"epoch": 1.401496,
"eval_test2_cer": 0.08579254862433269,
"eval_test2_cer_norm": 0.06814572358227455,
"eval_test2_loss": 0.3084600865840912,
"eval_test2_runtime": 2471.9908,
"eval_test2_samples_per_second": 1.011,
"eval_test2_steps_per_second": 0.253,
"eval_test2_wer": 0.19641222247653925,
"eval_test2_wer_norm": 0.13070135228054092,
"step": 112000
},
{
"epoch": 1.402296,
"grad_norm": 2.858332633972168,
"learning_rate": 1.04e-06,
"loss": 0.1044,
"step": 112100
},
{
"epoch": 1.4030960000000001,
"grad_norm": 5.488587379455566,
"learning_rate": 1.0319678714859437e-06,
"loss": 0.1366,
"step": 112200
},
{
"epoch": 1.403896,
"grad_norm": 13.426318168640137,
"learning_rate": 1.0239357429718876e-06,
"loss": 0.1186,
"step": 112300
},
{
"epoch": 1.404696,
"grad_norm": 4.668258190155029,
"learning_rate": 1.0159036144578315e-06,
"loss": 0.1056,
"step": 112400
},
{
"epoch": 1.405496,
"grad_norm": 3.497905969619751,
"learning_rate": 1.0078714859437753e-06,
"loss": 0.1168,
"step": 112500
},
{
"epoch": 1.406296,
"grad_norm": 5.060064315795898,
"learning_rate": 9.99839357429719e-07,
"loss": 0.1167,
"step": 112600
},
{
"epoch": 1.4070960000000001,
"grad_norm": 1.985062837600708,
"learning_rate": 9.918072289156627e-07,
"loss": 0.1234,
"step": 112700
},
{
"epoch": 1.407896,
"grad_norm": 3.2145867347717285,
"learning_rate": 9.837751004016064e-07,
"loss": 0.1231,
"step": 112800
},
{
"epoch": 1.408696,
"grad_norm": 6.194746971130371,
"learning_rate": 9.757429718875503e-07,
"loss": 0.1038,
"step": 112900
},
{
"epoch": 1.409496,
"grad_norm": 4.408815860748291,
"learning_rate": 9.67710843373494e-07,
"loss": 0.1157,
"step": 113000
},
{
"epoch": 1.410296,
"grad_norm": 4.587121486663818,
"learning_rate": 9.59678714859438e-07,
"loss": 0.1163,
"step": 113100
},
{
"epoch": 1.4110960000000001,
"grad_norm": 4.01561975479126,
"learning_rate": 9.516465863453816e-07,
"loss": 0.1077,
"step": 113200
},
{
"epoch": 1.411896,
"grad_norm": 4.253355026245117,
"learning_rate": 9.436144578313254e-07,
"loss": 0.1209,
"step": 113300
},
{
"epoch": 1.412696,
"grad_norm": 1.5479423999786377,
"learning_rate": 9.355823293172692e-07,
"loss": 0.1378,
"step": 113400
},
{
"epoch": 1.4134959999999999,
"grad_norm": 4.244668960571289,
"learning_rate": 9.27550200803213e-07,
"loss": 0.1381,
"step": 113500
},
{
"epoch": 1.414296,
"grad_norm": 4.061355113983154,
"learning_rate": 9.195180722891567e-07,
"loss": 0.1283,
"step": 113600
},
{
"epoch": 1.4150960000000001,
"grad_norm": 8.999344825744629,
"learning_rate": 9.114859437751005e-07,
"loss": 0.1291,
"step": 113700
},
{
"epoch": 1.415896,
"grad_norm": 3.263817310333252,
"learning_rate": 9.034538152610442e-07,
"loss": 0.119,
"step": 113800
},
{
"epoch": 1.416696,
"grad_norm": 5.394161224365234,
"learning_rate": 8.954216867469879e-07,
"loss": 0.1218,
"step": 113900
},
{
"epoch": 1.4174959999999999,
"grad_norm": 11.564488410949707,
"learning_rate": 8.873895582329318e-07,
"loss": 0.1222,
"step": 114000
},
{
"epoch": 1.418296,
"grad_norm": 3.848464250564575,
"learning_rate": 8.793574297188756e-07,
"loss": 0.1175,
"step": 114100
},
{
"epoch": 1.4190960000000001,
"grad_norm": 3.4740781784057617,
"learning_rate": 8.713253012048194e-07,
"loss": 0.1267,
"step": 114200
},
{
"epoch": 1.419896,
"grad_norm": 3.503969192504883,
"learning_rate": 8.632931726907632e-07,
"loss": 0.1333,
"step": 114300
},
{
"epoch": 1.420696,
"grad_norm": 3.658344268798828,
"learning_rate": 8.552610441767069e-07,
"loss": 0.1137,
"step": 114400
},
{
"epoch": 1.4214959999999999,
"grad_norm": 5.1479973793029785,
"learning_rate": 8.472289156626507e-07,
"loss": 0.1254,
"step": 114500
},
{
"epoch": 1.422296,
"grad_norm": 5.20347261428833,
"learning_rate": 8.391967871485944e-07,
"loss": 0.1167,
"step": 114600
},
{
"epoch": 1.4230960000000001,
"grad_norm": 5.197832107543945,
"learning_rate": 8.311646586345381e-07,
"loss": 0.1297,
"step": 114700
},
{
"epoch": 1.423896,
"grad_norm": 2.3857178688049316,
"learning_rate": 8.23132530120482e-07,
"loss": 0.1173,
"step": 114800
},
{
"epoch": 1.424696,
"grad_norm": 5.162271976470947,
"learning_rate": 8.151004016064258e-07,
"loss": 0.1089,
"step": 114900
},
{
"epoch": 1.4254959999999999,
"grad_norm": 4.035723686218262,
"learning_rate": 8.070682730923695e-07,
"loss": 0.1186,
"step": 115000
},
{
"epoch": 1.426296,
"grad_norm": 7.1610283851623535,
"learning_rate": 7.991164658634539e-07,
"loss": 0.1196,
"step": 115100
},
{
"epoch": 1.427096,
"grad_norm": 6.167757987976074,
"learning_rate": 7.910843373493976e-07,
"loss": 0.112,
"step": 115200
},
{
"epoch": 1.427896,
"grad_norm": 8.109100341796875,
"learning_rate": 7.830522088353415e-07,
"loss": 0.1092,
"step": 115300
},
{
"epoch": 1.428696,
"grad_norm": 2.2968735694885254,
"learning_rate": 7.750200803212853e-07,
"loss": 0.1058,
"step": 115400
},
{
"epoch": 1.4294959999999999,
"grad_norm": 4.020580768585205,
"learning_rate": 7.66987951807229e-07,
"loss": 0.1089,
"step": 115500
},
{
"epoch": 1.430296,
"grad_norm": 1.7853455543518066,
"learning_rate": 7.589558232931728e-07,
"loss": 0.1077,
"step": 115600
},
{
"epoch": 1.431096,
"grad_norm": 2.6213812828063965,
"learning_rate": 7.509236947791165e-07,
"loss": 0.114,
"step": 115700
},
{
"epoch": 1.431896,
"grad_norm": 7.169987678527832,
"learning_rate": 7.428915662650602e-07,
"loss": 0.1247,
"step": 115800
},
{
"epoch": 1.432696,
"grad_norm": 5.280975818634033,
"learning_rate": 7.348594377510041e-07,
"loss": 0.1117,
"step": 115900
},
{
"epoch": 1.4334959999999999,
"grad_norm": 7.396920204162598,
"learning_rate": 7.268273092369478e-07,
"loss": 0.1218,
"step": 116000
},
{
"epoch": 1.4334959999999999,
"eval_test1_cer": 0.04053700799163493,
"eval_test1_cer_norm": 0.02625399188416933,
"eval_test1_loss": 0.16978052258491516,
"eval_test1_runtime": 2459.5813,
"eval_test1_samples_per_second": 1.016,
"eval_test1_steps_per_second": 0.254,
"eval_test1_wer": 0.12650514592262166,
"eval_test1_wer_norm": 0.06905520324965662,
"step": 116000
},
{
"epoch": 1.4334959999999999,
"eval_test2_cer": 0.0906269832381379,
"eval_test2_cer_norm": 0.07077006507592191,
"eval_test2_loss": 0.30657845735549927,
"eval_test2_runtime": 2479.2829,
"eval_test2_samples_per_second": 1.008,
"eval_test2_steps_per_second": 0.252,
"eval_test2_wer": 0.20224879835202564,
"eval_test2_wer_norm": 0.13654595461838184,
"step": 116000
},
{
"epoch": 1.434296,
"grad_norm": 7.435978412628174,
"learning_rate": 7.187951807228916e-07,
"loss": 0.1312,
"step": 116100
},
{
"epoch": 1.435096,
"grad_norm": 6.736993789672852,
"learning_rate": 7.107630522088355e-07,
"loss": 0.1245,
"step": 116200
},
{
"epoch": 1.435896,
"grad_norm": 3.6509387493133545,
"learning_rate": 7.027309236947792e-07,
"loss": 0.1041,
"step": 116300
},
{
"epoch": 1.436696,
"grad_norm": 5.3854851722717285,
"learning_rate": 6.94698795180723e-07,
"loss": 0.126,
"step": 116400
},
{
"epoch": 1.4374959999999999,
"grad_norm": 5.283120155334473,
"learning_rate": 6.866666666666667e-07,
"loss": 0.1615,
"step": 116500
},
{
"epoch": 1.438296,
"grad_norm": 3.9473323822021484,
"learning_rate": 6.786345381526105e-07,
"loss": 0.1218,
"step": 116600
},
{
"epoch": 1.439096,
"grad_norm": 11.495280265808105,
"learning_rate": 6.706024096385542e-07,
"loss": 0.1165,
"step": 116700
},
{
"epoch": 1.439896,
"grad_norm": 6.106124401092529,
"learning_rate": 6.62570281124498e-07,
"loss": 0.1291,
"step": 116800
},
{
"epoch": 1.440696,
"grad_norm": 3.252533435821533,
"learning_rate": 6.545381526104418e-07,
"loss": 0.1302,
"step": 116900
},
{
"epoch": 1.4414959999999999,
"grad_norm": 4.417702674865723,
"learning_rate": 6.465060240963857e-07,
"loss": 0.1185,
"step": 117000
},
{
"epoch": 1.442296,
"grad_norm": 7.446925163269043,
"learning_rate": 6.384738955823294e-07,
"loss": 0.1181,
"step": 117100
},
{
"epoch": 1.443096,
"grad_norm": 4.945169448852539,
"learning_rate": 6.304417670682732e-07,
"loss": 0.1146,
"step": 117200
},
{
"epoch": 1.443896,
"grad_norm": 4.255674362182617,
"learning_rate": 6.224096385542169e-07,
"loss": 0.1149,
"step": 117300
},
{
"epoch": 1.444696,
"grad_norm": 7.243592739105225,
"learning_rate": 6.143775100401607e-07,
"loss": 0.1266,
"step": 117400
},
{
"epoch": 1.445496,
"grad_norm": 4.7151265144348145,
"learning_rate": 6.063453815261044e-07,
"loss": 0.1228,
"step": 117500
},
{
"epoch": 1.446296,
"grad_norm": 3.492438793182373,
"learning_rate": 5.983132530120483e-07,
"loss": 0.107,
"step": 117600
},
{
"epoch": 1.447096,
"grad_norm": 5.758606910705566,
"learning_rate": 5.902811244979921e-07,
"loss": 0.1104,
"step": 117700
},
{
"epoch": 1.447896,
"grad_norm": 5.847075462341309,
"learning_rate": 5.822489959839358e-07,
"loss": 0.1234,
"step": 117800
},
{
"epoch": 1.448696,
"grad_norm": 4.088864803314209,
"learning_rate": 5.742168674698796e-07,
"loss": 0.1247,
"step": 117900
},
{
"epoch": 1.449496,
"grad_norm": 2.7524282932281494,
"learning_rate": 5.661847389558233e-07,
"loss": 0.0971,
"step": 118000
},
{
"epoch": 1.450296,
"grad_norm": 6.989465236663818,
"learning_rate": 5.581526104417671e-07,
"loss": 0.126,
"step": 118100
},
{
"epoch": 1.451096,
"grad_norm": 9.567557334899902,
"learning_rate": 5.501204819277109e-07,
"loss": 0.1191,
"step": 118200
},
{
"epoch": 1.451896,
"grad_norm": 10.12435531616211,
"learning_rate": 5.420883534136546e-07,
"loss": 0.1096,
"step": 118300
},
{
"epoch": 1.452696,
"grad_norm": 1.7563780546188354,
"learning_rate": 5.340562248995985e-07,
"loss": 0.1197,
"step": 118400
},
{
"epoch": 1.453496,
"grad_norm": 5.218570232391357,
"learning_rate": 5.260240963855423e-07,
"loss": 0.1207,
"step": 118500
},
{
"epoch": 1.454296,
"grad_norm": 4.665302753448486,
"learning_rate": 5.17991967871486e-07,
"loss": 0.115,
"step": 118600
},
{
"epoch": 1.455096,
"grad_norm": 5.446451663970947,
"learning_rate": 5.099598393574298e-07,
"loss": 0.119,
"step": 118700
},
{
"epoch": 1.455896,
"grad_norm": 4.688202381134033,
"learning_rate": 5.020080321285141e-07,
"loss": 0.1369,
"step": 118800
},
{
"epoch": 1.456696,
"grad_norm": 3.5095245838165283,
"learning_rate": 4.939759036144578e-07,
"loss": 0.1094,
"step": 118900
},
{
"epoch": 1.457496,
"grad_norm": 3.919813394546509,
"learning_rate": 4.859437751004017e-07,
"loss": 0.1113,
"step": 119000
},
{
"epoch": 1.458296,
"grad_norm": 3.1427431106567383,
"learning_rate": 4.779116465863454e-07,
"loss": 0.1127,
"step": 119100
},
{
"epoch": 1.459096,
"grad_norm": 5.548600673675537,
"learning_rate": 4.698795180722892e-07,
"loss": 0.1175,
"step": 119200
},
{
"epoch": 1.459896,
"grad_norm": 3.4401326179504395,
"learning_rate": 4.6184738955823296e-07,
"loss": 0.1118,
"step": 119300
},
{
"epoch": 1.460696,
"grad_norm": 0.7513042688369751,
"learning_rate": 4.538152610441767e-07,
"loss": 0.1227,
"step": 119400
},
{
"epoch": 1.461496,
"grad_norm": 4.836828708648682,
"learning_rate": 4.4578313253012054e-07,
"loss": 0.1208,
"step": 119500
},
{
"epoch": 1.462296,
"grad_norm": 9.92773723602295,
"learning_rate": 4.377510040160643e-07,
"loss": 0.1021,
"step": 119600
},
{
"epoch": 1.463096,
"grad_norm": 4.6646575927734375,
"learning_rate": 4.2971887550200806e-07,
"loss": 0.1142,
"step": 119700
},
{
"epoch": 1.463896,
"grad_norm": 0.4946214258670807,
"learning_rate": 4.216867469879518e-07,
"loss": 0.1334,
"step": 119800
},
{
"epoch": 1.464696,
"grad_norm": 7.3409247398376465,
"learning_rate": 4.1365461847389564e-07,
"loss": 0.1387,
"step": 119900
},
{
"epoch": 1.465496,
"grad_norm": 2.3244385719299316,
"learning_rate": 4.056224899598394e-07,
"loss": 0.1216,
"step": 120000
},
{
"epoch": 1.465496,
"eval_test1_cer": 0.04472888191799238,
"eval_test1_cer_norm": 0.029865296419910196,
"eval_test1_loss": 0.16931886970996857,
"eval_test1_runtime": 2461.9854,
"eval_test1_samples_per_second": 1.015,
"eval_test1_steps_per_second": 0.254,
"eval_test1_wer": 0.13493104755240679,
"eval_test1_wer_norm": 0.076477980069552,
"step": 120000
},
{
"epoch": 1.465496,
"eval_test2_cer": 0.09805129353791019,
"eval_test2_cer_norm": 0.07766985590331578,
"eval_test2_loss": 0.3047462999820709,
"eval_test2_runtime": 2488.9087,
"eval_test2_samples_per_second": 1.004,
"eval_test2_steps_per_second": 0.251,
"eval_test2_wer": 0.21137560082398718,
"eval_test2_wer_norm": 0.1462296584918634,
"step": 120000
},
{
"epoch": 1.466296,
"grad_norm": 5.934924602508545,
"learning_rate": 3.9759036144578316e-07,
"loss": 0.1166,
"step": 120100
},
{
"epoch": 1.467096,
"grad_norm": 6.291118621826172,
"learning_rate": 3.895582329317269e-07,
"loss": 0.1144,
"step": 120200
},
{
"epoch": 1.467896,
"grad_norm": 7.283012866973877,
"learning_rate": 3.8152610441767073e-07,
"loss": 0.1258,
"step": 120300
},
{
"epoch": 1.468696,
"grad_norm": 2.082486391067505,
"learning_rate": 3.734939759036145e-07,
"loss": 0.1209,
"step": 120400
},
{
"epoch": 1.469496,
"grad_norm": 7.8560967445373535,
"learning_rate": 3.6546184738955826e-07,
"loss": 0.115,
"step": 120500
},
{
"epoch": 1.470296,
"grad_norm": 6.016510963439941,
"learning_rate": 3.57429718875502e-07,
"loss": 0.1082,
"step": 120600
},
{
"epoch": 1.471096,
"grad_norm": 2.1153485774993896,
"learning_rate": 3.4939759036144583e-07,
"loss": 0.1286,
"step": 120700
},
{
"epoch": 1.471896,
"grad_norm": 5.02449369430542,
"learning_rate": 3.413654618473896e-07,
"loss": 0.111,
"step": 120800
},
{
"epoch": 1.472696,
"grad_norm": 2.6848514080047607,
"learning_rate": 3.3333333333333335e-07,
"loss": 0.1161,
"step": 120900
},
{
"epoch": 1.473496,
"grad_norm": 3.306321620941162,
"learning_rate": 3.253012048192771e-07,
"loss": 0.1164,
"step": 121000
},
{
"epoch": 1.474296,
"grad_norm": 5.202853202819824,
"learning_rate": 3.1726907630522093e-07,
"loss": 0.1148,
"step": 121100
},
{
"epoch": 1.475096,
"grad_norm": 6.273738861083984,
"learning_rate": 3.092369477911647e-07,
"loss": 0.1257,
"step": 121200
},
{
"epoch": 1.475896,
"grad_norm": 1.6317389011383057,
"learning_rate": 3.01285140562249e-07,
"loss": 0.1169,
"step": 121300
},
{
"epoch": 1.476696,
"grad_norm": 3.1588313579559326,
"learning_rate": 2.932530120481928e-07,
"loss": 0.1243,
"step": 121400
},
{
"epoch": 1.477496,
"grad_norm": 3.1574482917785645,
"learning_rate": 2.852208835341366e-07,
"loss": 0.1254,
"step": 121500
},
{
"epoch": 1.478296,
"grad_norm": 3.313225507736206,
"learning_rate": 2.7718875502008034e-07,
"loss": 0.1196,
"step": 121600
},
{
"epoch": 1.479096,
"grad_norm": 3.645125389099121,
"learning_rate": 2.691566265060241e-07,
"loss": 0.0956,
"step": 121700
},
{
"epoch": 1.479896,
"grad_norm": 0.8086249232292175,
"learning_rate": 2.611244979919679e-07,
"loss": 0.1171,
"step": 121800
},
{
"epoch": 1.480696,
"grad_norm": 4.813057899475098,
"learning_rate": 2.530923694779117e-07,
"loss": 0.1254,
"step": 121900
},
{
"epoch": 1.481496,
"grad_norm": 3.0106194019317627,
"learning_rate": 2.4506024096385544e-07,
"loss": 0.109,
"step": 122000
},
{
"epoch": 1.482296,
"grad_norm": 3.1917545795440674,
"learning_rate": 2.370281124497992e-07,
"loss": 0.1082,
"step": 122100
},
{
"epoch": 1.483096,
"grad_norm": 8.249537467956543,
"learning_rate": 2.28995983935743e-07,
"loss": 0.1334,
"step": 122200
},
{
"epoch": 1.483896,
"grad_norm": 2.7427356243133545,
"learning_rate": 2.2096385542168677e-07,
"loss": 0.1209,
"step": 122300
},
{
"epoch": 1.484696,
"grad_norm": 2.4920859336853027,
"learning_rate": 2.1293172690763056e-07,
"loss": 0.1041,
"step": 122400
},
{
"epoch": 1.485496,
"grad_norm": 5.013920307159424,
"learning_rate": 2.0489959839357432e-07,
"loss": 0.1114,
"step": 122500
},
{
"epoch": 1.486296,
"grad_norm": 0.5811383128166199,
"learning_rate": 1.968674698795181e-07,
"loss": 0.1228,
"step": 122600
},
{
"epoch": 1.487096,
"grad_norm": 5.773433685302734,
"learning_rate": 1.8883534136546187e-07,
"loss": 0.1284,
"step": 122700
},
{
"epoch": 1.487896,
"grad_norm": 3.8643128871917725,
"learning_rate": 1.8080321285140566e-07,
"loss": 0.1159,
"step": 122800
},
{
"epoch": 1.488696,
"grad_norm": 5.357062816619873,
"learning_rate": 1.7277108433734942e-07,
"loss": 0.1225,
"step": 122900
},
{
"epoch": 1.489496,
"grad_norm": 4.911788463592529,
"learning_rate": 1.647389558232932e-07,
"loss": 0.1116,
"step": 123000
},
{
"epoch": 1.490296,
"grad_norm": 4.041905403137207,
"learning_rate": 1.5670682730923697e-07,
"loss": 0.1236,
"step": 123100
},
{
"epoch": 1.491096,
"grad_norm": 5.671024799346924,
"learning_rate": 1.4867469879518073e-07,
"loss": 0.1168,
"step": 123200
},
{
"epoch": 1.491896,
"grad_norm": 8.380229949951172,
"learning_rate": 1.4064257028112452e-07,
"loss": 0.1192,
"step": 123300
},
{
"epoch": 1.492696,
"grad_norm": 8.371071815490723,
"learning_rate": 1.3261044176706828e-07,
"loss": 0.102,
"step": 123400
},
{
"epoch": 1.493496,
"grad_norm": 4.045628547668457,
"learning_rate": 1.2457831325301207e-07,
"loss": 0.1177,
"step": 123500
},
{
"epoch": 1.494296,
"grad_norm": 7.542957782745361,
"learning_rate": 1.1654618473895584e-07,
"loss": 0.108,
"step": 123600
},
{
"epoch": 2.000592,
"grad_norm": 5.552263259887695,
"learning_rate": 1.0851405622489961e-07,
"loss": 0.1093,
"step": 123700
},
{
"epoch": 2.001392,
"grad_norm": 4.822888374328613,
"learning_rate": 1.0048192771084339e-07,
"loss": 0.1042,
"step": 123800
},
{
"epoch": 2.002192,
"grad_norm": 0.32907435297966003,
"learning_rate": 9.244979919678716e-08,
"loss": 0.1014,
"step": 123900
},
{
"epoch": 2.002992,
"grad_norm": 5.526859760284424,
"learning_rate": 8.441767068273094e-08,
"loss": 0.0819,
"step": 124000
},
{
"epoch": 2.002992,
"eval_test1_cer": 0.04037829561580402,
"eval_test1_cer_norm": 0.026181957884121306,
"eval_test1_loss": 0.1684691458940506,
"eval_test1_runtime": 2457.6805,
"eval_test1_samples_per_second": 1.017,
"eval_test1_steps_per_second": 0.254,
"eval_test1_wer": 0.1262135922330097,
"eval_test1_wer_norm": 0.06817849732604693,
"step": 124000
},
{
"epoch": 2.002992,
"eval_test2_cer": 0.09239089857020197,
"eval_test2_cer_norm": 0.07192245119305857,
"eval_test2_loss": 0.30393916368484497,
"eval_test2_runtime": 2485.4178,
"eval_test2_samples_per_second": 1.006,
"eval_test2_steps_per_second": 0.251,
"eval_test2_wer": 0.20482375829709315,
"eval_test2_wer_norm": 0.13981205592482238,
"step": 124000
}
],
"logging_steps": 100,
"max_steps": 125000,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 4000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 5.062139105004749e+20,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}