{
"best_global_step": 4000,
"best_metric": 0.2779,
"best_model_checkpoint": "/tmp/skiredj.abderrahman_Bstar9train/out/output/checkpoint-4000",
"epoch": 3.0,
"eval_steps": 1000,
"global_step": 6741,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.022256843979523704,
"grad_norm": 1.5087908506393433,
"learning_rate": 4.9000000000000005e-06,
"loss": 0.1324,
"step": 50
},
{
"epoch": 0.04451368795904741,
"grad_norm": 1.7243016958236694,
"learning_rate": 9.9e-06,
"loss": 0.1312,
"step": 100
},
{
"epoch": 0.0667705319385711,
"grad_norm": 1.633836269378662,
"learning_rate": 9.926215931335644e-06,
"loss": 0.1448,
"step": 150
},
{
"epoch": 0.08902737591809481,
"grad_norm": 1.470070719718933,
"learning_rate": 9.850926065351604e-06,
"loss": 0.1475,
"step": 200
},
{
"epoch": 0.11128421989761851,
"grad_norm": 1.4569244384765625,
"learning_rate": 9.775636199367566e-06,
"loss": 0.1453,
"step": 250
},
{
"epoch": 0.1335410638771422,
"grad_norm": 1.2694658041000366,
"learning_rate": 9.700346333383527e-06,
"loss": 0.1448,
"step": 300
},
{
"epoch": 0.15579790785666592,
"grad_norm": 1.5997624397277832,
"learning_rate": 9.625056467399489e-06,
"loss": 0.1483,
"step": 350
},
{
"epoch": 0.17805475183618963,
"grad_norm": 1.787446141242981,
"learning_rate": 9.54976660141545e-06,
"loss": 0.1493,
"step": 400
},
{
"epoch": 0.20031159581571334,
"grad_norm": 1.745363473892212,
"learning_rate": 9.474476735431411e-06,
"loss": 0.1518,
"step": 450
},
{
"epoch": 0.22256843979523702,
"grad_norm": 1.36530601978302,
"learning_rate": 9.399186869447373e-06,
"loss": 0.1378,
"step": 500
},
{
"epoch": 0.24482528377476073,
"grad_norm": 1.6621060371398926,
"learning_rate": 9.323897003463336e-06,
"loss": 0.1433,
"step": 550
},
{
"epoch": 0.2670821277542844,
"grad_norm": 1.249980092048645,
"learning_rate": 9.248607137479296e-06,
"loss": 0.1405,
"step": 600
},
{
"epoch": 0.2893389717338081,
"grad_norm": 1.4200546741485596,
"learning_rate": 9.173317271495258e-06,
"loss": 0.146,
"step": 650
},
{
"epoch": 0.31159581571333184,
"grad_norm": 1.2627590894699097,
"learning_rate": 9.09802740551122e-06,
"loss": 0.1481,
"step": 700
},
{
"epoch": 0.33385265969285555,
"grad_norm": 1.6646721363067627,
"learning_rate": 9.02273753952718e-06,
"loss": 0.1425,
"step": 750
},
{
"epoch": 0.35610950367237926,
"grad_norm": 1.3609075546264648,
"learning_rate": 8.947447673543141e-06,
"loss": 0.1414,
"step": 800
},
{
"epoch": 0.37836634765190297,
"grad_norm": 1.5529732704162598,
"learning_rate": 8.872157807559103e-06,
"loss": 0.1455,
"step": 850
},
{
"epoch": 0.4006231916314267,
"grad_norm": 1.4032143354415894,
"learning_rate": 8.796867941575065e-06,
"loss": 0.1451,
"step": 900
},
{
"epoch": 0.4228800356109504,
"grad_norm": 1.1964385509490967,
"learning_rate": 8.721578075591026e-06,
"loss": 0.1425,
"step": 950
},
{
"epoch": 0.44513687959047404,
"grad_norm": 1.4103286266326904,
"learning_rate": 8.646288209606988e-06,
"loss": 0.1406,
"step": 1000
},
{
"epoch": 0.44513687959047404,
"eval_cer": 0.2041,
"eval_loss": 0.47862356901168823,
"eval_mer": 0.2497,
"eval_runtime": 531.5273,
"eval_samples_per_second": 5.006,
"eval_steps_per_second": 0.181,
"eval_wer": 0.2893,
"eval_wil": 0.3493,
"step": 1000
},
{
"epoch": 0.46739372356999775,
"grad_norm": 1.302578330039978,
"learning_rate": 8.57099834362295e-06,
"loss": 0.1455,
"step": 1050
},
{
"epoch": 0.48965056754952146,
"grad_norm": 2.0349037647247314,
"learning_rate": 8.49570847763891e-06,
"loss": 0.136,
"step": 1100
},
{
"epoch": 0.5119074115290452,
"grad_norm": 1.5251567363739014,
"learning_rate": 8.42041861165487e-06,
"loss": 0.1454,
"step": 1150
},
{
"epoch": 0.5341642555085688,
"grad_norm": 1.3429796695709229,
"learning_rate": 8.345128745670833e-06,
"loss": 0.1413,
"step": 1200
},
{
"epoch": 0.5564210994880926,
"grad_norm": 1.482814073562622,
"learning_rate": 8.269838879686795e-06,
"loss": 0.1391,
"step": 1250
},
{
"epoch": 0.5786779434676163,
"grad_norm": 1.428371548652649,
"learning_rate": 8.194549013702757e-06,
"loss": 0.1381,
"step": 1300
},
{
"epoch": 0.60093478744714,
"grad_norm": 1.4925706386566162,
"learning_rate": 8.119259147718717e-06,
"loss": 0.1407,
"step": 1350
},
{
"epoch": 0.6231916314266637,
"grad_norm": 1.4694454669952393,
"learning_rate": 8.04396928173468e-06,
"loss": 0.1402,
"step": 1400
},
{
"epoch": 0.6454484754061874,
"grad_norm": 1.3397713899612427,
"learning_rate": 7.968679415750641e-06,
"loss": 0.1351,
"step": 1450
},
{
"epoch": 0.6677053193857111,
"grad_norm": 1.4111981391906738,
"learning_rate": 7.893389549766602e-06,
"loss": 0.1393,
"step": 1500
},
{
"epoch": 0.6899621633652349,
"grad_norm": 1.5216033458709717,
"learning_rate": 7.818099683782564e-06,
"loss": 0.1375,
"step": 1550
},
{
"epoch": 0.7122190073447585,
"grad_norm": 1.6828105449676514,
"learning_rate": 7.742809817798524e-06,
"loss": 0.1354,
"step": 1600
},
{
"epoch": 0.7344758513242822,
"grad_norm": 1.4353023767471313,
"learning_rate": 7.667519951814486e-06,
"loss": 0.1382,
"step": 1650
},
{
"epoch": 0.7567326953038059,
"grad_norm": 1.4738577604293823,
"learning_rate": 7.592230085830448e-06,
"loss": 0.1366,
"step": 1700
},
{
"epoch": 0.7789895392833296,
"grad_norm": 1.192264199256897,
"learning_rate": 7.516940219846409e-06,
"loss": 0.1389,
"step": 1750
},
{
"epoch": 0.8012463832628534,
"grad_norm": 1.484379768371582,
"learning_rate": 7.441650353862371e-06,
"loss": 0.135,
"step": 1800
},
{
"epoch": 0.823503227242377,
"grad_norm": 1.3312201499938965,
"learning_rate": 7.366360487878332e-06,
"loss": 0.1318,
"step": 1850
},
{
"epoch": 0.8457600712219008,
"grad_norm": 1.2523818016052246,
"learning_rate": 7.2910706218942936e-06,
"loss": 0.1428,
"step": 1900
},
{
"epoch": 0.8680169152014244,
"grad_norm": 1.3878048658370972,
"learning_rate": 7.215780755910256e-06,
"loss": 0.1369,
"step": 1950
},
{
"epoch": 0.8902737591809481,
"grad_norm": 1.4441479444503784,
"learning_rate": 7.140490889926216e-06,
"loss": 0.1396,
"step": 2000
},
{
"epoch": 0.8902737591809481,
"eval_cer": 0.197,
"eval_loss": 0.4693019390106201,
"eval_mer": 0.2448,
"eval_runtime": 532.3783,
"eval_samples_per_second": 4.998,
"eval_steps_per_second": 0.18,
"eval_wer": 0.2837,
"eval_wil": 0.3432,
"step": 2000
},
{
"epoch": 0.9125306031604719,
"grad_norm": 1.4133597612380981,
"learning_rate": 7.065201023942177e-06,
"loss": 0.1368,
"step": 2050
},
{
"epoch": 0.9347874471399955,
"grad_norm": 1.5449390411376953,
"learning_rate": 6.9899111579581394e-06,
"loss": 0.1379,
"step": 2100
},
{
"epoch": 0.9570442911195193,
"grad_norm": 1.6329177618026733,
"learning_rate": 6.914621291974101e-06,
"loss": 0.1364,
"step": 2150
},
{
"epoch": 0.9793011350990429,
"grad_norm": 1.6076219081878662,
"learning_rate": 6.839331425990062e-06,
"loss": 0.1365,
"step": 2200
},
{
"epoch": 1.0013354106387715,
"grad_norm": 0.9237516522407532,
"learning_rate": 6.764041560006024e-06,
"loss": 0.1314,
"step": 2250
},
{
"epoch": 1.0235922546182952,
"grad_norm": 1.2924916744232178,
"learning_rate": 6.688751694021985e-06,
"loss": 0.0808,
"step": 2300
},
{
"epoch": 1.0458490985978188,
"grad_norm": 0.9751130938529968,
"learning_rate": 6.613461828037947e-06,
"loss": 0.0785,
"step": 2350
},
{
"epoch": 1.0681059425773425,
"grad_norm": 1.2038739919662476,
"learning_rate": 6.538171962053909e-06,
"loss": 0.0787,
"step": 2400
},
{
"epoch": 1.0903627865568661,
"grad_norm": 1.1433651447296143,
"learning_rate": 6.462882096069869e-06,
"loss": 0.0738,
"step": 2450
},
{
"epoch": 1.11261963053639,
"grad_norm": 1.1120831966400146,
"learning_rate": 6.38759223008583e-06,
"loss": 0.0748,
"step": 2500
},
{
"epoch": 1.1348764745159137,
"grad_norm": 1.209842324256897,
"learning_rate": 6.312302364101792e-06,
"loss": 0.079,
"step": 2550
},
{
"epoch": 1.1571333184954373,
"grad_norm": 1.3221111297607422,
"learning_rate": 6.237012498117754e-06,
"loss": 0.0797,
"step": 2600
},
{
"epoch": 1.179390162474961,
"grad_norm": 1.1365020275115967,
"learning_rate": 6.161722632133716e-06,
"loss": 0.0779,
"step": 2650
},
{
"epoch": 1.2016470064544849,
"grad_norm": 1.0401190519332886,
"learning_rate": 6.086432766149677e-06,
"loss": 0.0774,
"step": 2700
},
{
"epoch": 1.2239038504340085,
"grad_norm": 1.1138839721679688,
"learning_rate": 6.011142900165638e-06,
"loss": 0.0794,
"step": 2750
},
{
"epoch": 1.2461606944135322,
"grad_norm": 1.2989375591278076,
"learning_rate": 5.9358530341816e-06,
"loss": 0.0774,
"step": 2800
},
{
"epoch": 1.2684175383930558,
"grad_norm": 1.1034380197525024,
"learning_rate": 5.860563168197561e-06,
"loss": 0.0791,
"step": 2850
},
{
"epoch": 1.2906743823725795,
"grad_norm": 1.1647659540176392,
"learning_rate": 5.785273302213522e-06,
"loss": 0.0775,
"step": 2900
},
{
"epoch": 1.3129312263521034,
"grad_norm": 1.4231784343719482,
"learning_rate": 5.709983436229484e-06,
"loss": 0.0779,
"step": 2950
},
{
"epoch": 1.335188070331627,
"grad_norm": 1.1819056272506714,
"learning_rate": 5.634693570245445e-06,
"loss": 0.0836,
"step": 3000
},
{
"epoch": 1.335188070331627,
"eval_cer": 0.1976,
"eval_loss": 0.5102627277374268,
"eval_mer": 0.2454,
"eval_runtime": 524.3267,
"eval_samples_per_second": 5.075,
"eval_steps_per_second": 0.183,
"eval_wer": 0.2862,
"eval_wil": 0.344,
"step": 3000
},
{
"epoch": 1.3574449143111507,
"grad_norm": 1.1431699991226196,
"learning_rate": 5.559403704261407e-06,
"loss": 0.0783,
"step": 3050
},
{
"epoch": 1.3797017582906743,
"grad_norm": 1.000010371208191,
"learning_rate": 5.484113838277369e-06,
"loss": 0.0742,
"step": 3100
},
{
"epoch": 1.4019586022701982,
"grad_norm": 1.2086879014968872,
"learning_rate": 5.40882397229333e-06,
"loss": 0.077,
"step": 3150
},
{
"epoch": 1.4242154462497219,
"grad_norm": 1.0167688131332397,
"learning_rate": 5.333534106309291e-06,
"loss": 0.0782,
"step": 3200
},
{
"epoch": 1.4464722902292455,
"grad_norm": 1.3445148468017578,
"learning_rate": 5.258244240325253e-06,
"loss": 0.0754,
"step": 3250
},
{
"epoch": 1.4687291342087692,
"grad_norm": 1.6231786012649536,
"learning_rate": 5.182954374341214e-06,
"loss": 0.08,
"step": 3300
},
{
"epoch": 1.4909859781882928,
"grad_norm": 1.2671375274658203,
"learning_rate": 5.107664508357175e-06,
"loss": 0.076,
"step": 3350
},
{
"epoch": 1.5132428221678165,
"grad_norm": 1.6796362400054932,
"learning_rate": 5.032374642373137e-06,
"loss": 0.074,
"step": 3400
},
{
"epoch": 1.5354996661473403,
"grad_norm": 1.1336886882781982,
"learning_rate": 4.957084776389098e-06,
"loss": 0.0763,
"step": 3450
},
{
"epoch": 1.557756510126864,
"grad_norm": 1.1634091138839722,
"learning_rate": 4.88179491040506e-06,
"loss": 0.0794,
"step": 3500
},
{
"epoch": 1.5800133541063879,
"grad_norm": 1.4456815719604492,
"learning_rate": 4.806505044421022e-06,
"loss": 0.0778,
"step": 3550
},
{
"epoch": 1.6022701980859115,
"grad_norm": 1.084285855293274,
"learning_rate": 4.731215178436983e-06,
"loss": 0.0782,
"step": 3600
},
{
"epoch": 1.6245270420654352,
"grad_norm": 1.304287075996399,
"learning_rate": 4.655925312452944e-06,
"loss": 0.0764,
"step": 3650
},
{
"epoch": 1.6467838860449588,
"grad_norm": 1.10267972946167,
"learning_rate": 4.5806354464689054e-06,
"loss": 0.079,
"step": 3700
},
{
"epoch": 1.6690407300244825,
"grad_norm": 1.3505820035934448,
"learning_rate": 4.5053455804848675e-06,
"loss": 0.0752,
"step": 3750
},
{
"epoch": 1.6912975740040062,
"grad_norm": 1.0025562047958374,
"learning_rate": 4.430055714500828e-06,
"loss": 0.0748,
"step": 3800
},
{
"epoch": 1.7135544179835298,
"grad_norm": 0.9700995683670044,
"learning_rate": 4.35476584851679e-06,
"loss": 0.0769,
"step": 3850
},
{
"epoch": 1.7358112619630537,
"grad_norm": 1.2612080574035645,
"learning_rate": 4.279475982532751e-06,
"loss": 0.0753,
"step": 3900
},
{
"epoch": 1.7580681059425773,
"grad_norm": 1.264329433441162,
"learning_rate": 4.204186116548713e-06,
"loss": 0.0745,
"step": 3950
},
{
"epoch": 1.780324949922101,
"grad_norm": 1.1244704723358154,
"learning_rate": 4.128896250564675e-06,
"loss": 0.0727,
"step": 4000
},
{
"epoch": 1.780324949922101,
"eval_cer": 0.1927,
"eval_loss": 0.5105198621749878,
"eval_mer": 0.2404,
"eval_runtime": 525.2269,
"eval_samples_per_second": 5.066,
"eval_steps_per_second": 0.183,
"eval_wer": 0.2779,
"eval_wil": 0.337,
"step": 4000
},
{
"epoch": 1.8025817939016249,
"grad_norm": 1.4443764686584473,
"learning_rate": 4.053606384580636e-06,
"loss": 0.0767,
"step": 4050
},
{
"epoch": 1.8248386378811485,
"grad_norm": 1.1450945138931274,
"learning_rate": 3.978316518596597e-06,
"loss": 0.0731,
"step": 4100
},
{
"epoch": 1.8470954818606722,
"grad_norm": 1.2433844804763794,
"learning_rate": 3.903026652612558e-06,
"loss": 0.0786,
"step": 4150
},
{
"epoch": 1.8693523258401958,
"grad_norm": 0.9511911273002625,
"learning_rate": 3.8277367866285205e-06,
"loss": 0.0756,
"step": 4200
},
{
"epoch": 1.8916091698197195,
"grad_norm": 1.3445976972579956,
"learning_rate": 3.7524469206444813e-06,
"loss": 0.0767,
"step": 4250
},
{
"epoch": 1.9138660137992431,
"grad_norm": 1.1332846879959106,
"learning_rate": 3.677157054660443e-06,
"loss": 0.0731,
"step": 4300
},
{
"epoch": 1.936122857778767,
"grad_norm": 1.0032987594604492,
"learning_rate": 3.6018671886764047e-06,
"loss": 0.0743,
"step": 4350
},
{
"epoch": 1.9583797017582907,
"grad_norm": 1.3232260942459106,
"learning_rate": 3.526577322692366e-06,
"loss": 0.0741,
"step": 4400
},
{
"epoch": 1.9806365457378143,
"grad_norm": 1.2991896867752075,
"learning_rate": 3.451287456708327e-06,
"loss": 0.0749,
"step": 4450
},
{
"epoch": 2.002670821277543,
"grad_norm": 0.7459942698478699,
"learning_rate": 3.375997590724289e-06,
"loss": 0.0695,
"step": 4500
},
{
"epoch": 2.0249276652570667,
"grad_norm": 0.967068612575531,
"learning_rate": 3.30070772474025e-06,
"loss": 0.0408,
"step": 4550
},
{
"epoch": 2.0471845092365903,
"grad_norm": 1.0247790813446045,
"learning_rate": 3.225417858756212e-06,
"loss": 0.0379,
"step": 4600
},
{
"epoch": 2.069441353216114,
"grad_norm": 0.96858811378479,
"learning_rate": 3.150127992772173e-06,
"loss": 0.0387,
"step": 4650
},
{
"epoch": 2.0916981971956377,
"grad_norm": 1.1326206922531128,
"learning_rate": 3.0748381267881343e-06,
"loss": 0.0427,
"step": 4700
},
{
"epoch": 2.1139550411751613,
"grad_norm": 0.8742848634719849,
"learning_rate": 2.999548260804096e-06,
"loss": 0.0379,
"step": 4750
},
{
"epoch": 2.136211885154685,
"grad_norm": 0.7699297070503235,
"learning_rate": 2.9242583948200577e-06,
"loss": 0.04,
"step": 4800
},
{
"epoch": 2.1584687291342086,
"grad_norm": 0.8700041174888611,
"learning_rate": 2.8489685288360193e-06,
"loss": 0.0403,
"step": 4850
},
{
"epoch": 2.1807255731137323,
"grad_norm": 0.8692020773887634,
"learning_rate": 2.77367866285198e-06,
"loss": 0.0406,
"step": 4900
},
{
"epoch": 2.2029824170932564,
"grad_norm": 0.9844094514846802,
"learning_rate": 2.698388796867942e-06,
"loss": 0.0389,
"step": 4950
},
{
"epoch": 2.22523926107278,
"grad_norm": 0.7846298217773438,
"learning_rate": 2.6230989308839035e-06,
"loss": 0.0413,
"step": 5000
},
{
"epoch": 2.22523926107278,
"eval_cer": 0.1971,
"eval_loss": 0.5635275840759277,
"eval_mer": 0.2437,
"eval_runtime": 529.1595,
"eval_samples_per_second": 5.029,
"eval_steps_per_second": 0.181,
"eval_wer": 0.2842,
"eval_wil": 0.3413,
"step": 5000
},
{
"epoch": 2.2474961050523037,
"grad_norm": 0.8958929181098938,
"learning_rate": 2.5478090648998648e-06,
"loss": 0.0393,
"step": 5050
},
{
"epoch": 2.2697529490318273,
"grad_norm": 0.9802588224411011,
"learning_rate": 2.472519198915826e-06,
"loss": 0.0408,
"step": 5100
},
{
"epoch": 2.292009793011351,
"grad_norm": 1.052999496459961,
"learning_rate": 2.3972293329317877e-06,
"loss": 0.0399,
"step": 5150
},
{
"epoch": 2.3142666369908746,
"grad_norm": 0.6849333047866821,
"learning_rate": 2.321939466947749e-06,
"loss": 0.0388,
"step": 5200
},
{
"epoch": 2.3365234809703983,
"grad_norm": 1.0287905931472778,
"learning_rate": 2.24664960096371e-06,
"loss": 0.0408,
"step": 5250
},
{
"epoch": 2.358780324949922,
"grad_norm": 0.8322626352310181,
"learning_rate": 2.171359734979672e-06,
"loss": 0.0395,
"step": 5300
},
{
"epoch": 2.3810371689294456,
"grad_norm": 0.9950063228607178,
"learning_rate": 2.096069868995633e-06,
"loss": 0.0376,
"step": 5350
},
{
"epoch": 2.4032940129089697,
"grad_norm": 1.0962845087051392,
"learning_rate": 2.020780003011595e-06,
"loss": 0.0412,
"step": 5400
},
{
"epoch": 2.4255508568884934,
"grad_norm": 0.7699467539787292,
"learning_rate": 1.945490137027556e-06,
"loss": 0.0396,
"step": 5450
},
{
"epoch": 2.447807700868017,
"grad_norm": 1.1390819549560547,
"learning_rate": 1.8702002710435177e-06,
"loss": 0.0395,
"step": 5500
},
{
"epoch": 2.4700645448475407,
"grad_norm": 0.8535729646682739,
"learning_rate": 1.7949104050594792e-06,
"loss": 0.04,
"step": 5550
},
{
"epoch": 2.4923213888270643,
"grad_norm": 0.9253916144371033,
"learning_rate": 1.7196205390754407e-06,
"loss": 0.0399,
"step": 5600
},
{
"epoch": 2.514578232806588,
"grad_norm": 0.9185721278190613,
"learning_rate": 1.6443306730914021e-06,
"loss": 0.037,
"step": 5650
},
{
"epoch": 2.5368350767861116,
"grad_norm": 0.7001262307167053,
"learning_rate": 1.5690408071073634e-06,
"loss": 0.0398,
"step": 5700
},
{
"epoch": 2.5590919207656353,
"grad_norm": 0.869707465171814,
"learning_rate": 1.493750941123325e-06,
"loss": 0.0371,
"step": 5750
},
{
"epoch": 2.581348764745159,
"grad_norm": 1.1500749588012695,
"learning_rate": 1.4184610751392863e-06,
"loss": 0.0403,
"step": 5800
},
{
"epoch": 2.603605608724683,
"grad_norm": 0.7804761528968811,
"learning_rate": 1.343171209155248e-06,
"loss": 0.0403,
"step": 5850
},
{
"epoch": 2.6258624527042067,
"grad_norm": 0.7209272384643555,
"learning_rate": 1.2678813431712092e-06,
"loss": 0.0371,
"step": 5900
},
{
"epoch": 2.6481192966837304,
"grad_norm": 1.2576780319213867,
"learning_rate": 1.1925914771871707e-06,
"loss": 0.0381,
"step": 5950
},
{
"epoch": 2.670376140663254,
"grad_norm": 1.1125082969665527,
"learning_rate": 1.1173016112031322e-06,
"loss": 0.0366,
"step": 6000
},
{
"epoch": 2.670376140663254,
"eval_cer": 0.1994,
"eval_loss": 0.5712248682975769,
"eval_mer": 0.2444,
"eval_runtime": 528.5867,
"eval_samples_per_second": 5.034,
"eval_steps_per_second": 0.182,
"eval_wer": 0.2862,
"eval_wil": 0.3426,
"step": 6000
},
{
"epoch": 2.6926329846427777,
"grad_norm": 1.0231956243515015,
"learning_rate": 1.0420117452190936e-06,
"loss": 0.0396,
"step": 6050
},
{
"epoch": 2.7148898286223013,
"grad_norm": 0.9566596746444702,
"learning_rate": 9.66721879235055e-07,
"loss": 0.0355,
"step": 6100
},
{
"epoch": 2.737146672601825,
"grad_norm": 1.1378464698791504,
"learning_rate": 8.914320132510166e-07,
"loss": 0.0374,
"step": 6150
},
{
"epoch": 2.7594035165813486,
"grad_norm": 0.9265817999839783,
"learning_rate": 8.16142147266978e-07,
"loss": 0.0386,
"step": 6200
},
{
"epoch": 2.7816603605608723,
"grad_norm": 0.921313464641571,
"learning_rate": 7.408522812829394e-07,
"loss": 0.037,
"step": 6250
},
{
"epoch": 2.8039172045403964,
"grad_norm": 0.7966319918632507,
"learning_rate": 6.655624152989008e-07,
"loss": 0.0383,
"step": 6300
},
{
"epoch": 2.8261740485199196,
"grad_norm": 0.9206870794296265,
"learning_rate": 5.902725493148623e-07,
"loss": 0.0377,
"step": 6350
},
{
"epoch": 2.8484308924994437,
"grad_norm": 1.0037342309951782,
"learning_rate": 5.149826833308237e-07,
"loss": 0.0365,
"step": 6400
},
{
"epoch": 2.8706877364789674,
"grad_norm": 0.8986579775810242,
"learning_rate": 4.396928173467851e-07,
"loss": 0.0373,
"step": 6450
},
{
"epoch": 2.892944580458491,
"grad_norm": 0.9311175346374512,
"learning_rate": 3.644029513627466e-07,
"loss": 0.0397,
"step": 6500
},
{
"epoch": 2.9152014244380147,
"grad_norm": 1.158396601676941,
"learning_rate": 2.8911308537870805e-07,
"loss": 0.0368,
"step": 6550
},
{
"epoch": 2.9374582684175383,
"grad_norm": 0.8478817343711853,
"learning_rate": 2.138232193946695e-07,
"loss": 0.0365,
"step": 6600
},
{
"epoch": 2.959715112397062,
"grad_norm": 0.7029297351837158,
"learning_rate": 1.3853335341063093e-07,
"loss": 0.0371,
"step": 6650
},
{
"epoch": 2.9819719563765856,
"grad_norm": 0.8521508574485779,
"learning_rate": 6.324348742659239e-08,
"loss": 0.0352,
"step": 6700
},
{
"epoch": 3.0,
"eval_cer": 0.1974,
"eval_loss": 0.5691649317741394,
"eval_mer": 0.2424,
"eval_runtime": 519.6237,
"eval_samples_per_second": 5.121,
"eval_steps_per_second": 0.185,
"eval_wer": 0.2835,
"eval_wil": 0.3395,
"step": 6741
},
{
"epoch": 3.0,
"step": 6741,
"total_flos": 6.411282576614527e+20,
"train_loss": 0.08534082535589649,
"train_runtime": 19809.0009,
"train_samples_per_second": 9.525,
"train_steps_per_second": 0.34
}
],
"logging_steps": 50,
"max_steps": 6741,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 1000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 6.411282576614527e+20,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}
{
"best_global_step": 4000,
"best_metric": 0.2779,
"best_model_checkpoint": "/tmp/skiredj.abderrahman_Bstar9train/out/output/checkpoint-4000",
"epoch": 3.0,
"eval_steps": 1000,
"global_step": 6741,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.022256843979523704,
"grad_norm": 1.5087908506393433,
"learning_rate": 4.9000000000000005e-06,
"loss": 0.1324,
"step": 50
},
{
"epoch": 0.04451368795904741,
"grad_norm": 1.7243016958236694,
"learning_rate": 9.9e-06,
"loss": 0.1312,
"step": 100
},
{
"epoch": 0.0667705319385711,
"grad_norm": 1.633836269378662,
"learning_rate": 9.926215931335644e-06,
"loss": 0.1448,
"step": 150
},
{
"epoch": 0.08902737591809481,
"grad_norm": 1.470070719718933,
"learning_rate": 9.850926065351604e-06,
"loss": 0.1475,
"step": 200
},
{
"epoch": 0.11128421989761851,
"grad_norm": 1.4569244384765625,
"learning_rate": 9.775636199367566e-06,
"loss": 0.1453,
"step": 250
},
{
"epoch": 0.1335410638771422,
"grad_norm": 1.2694658041000366,
"learning_rate": 9.700346333383527e-06,
"loss": 0.1448,
"step": 300
},
{
"epoch": 0.15579790785666592,
"grad_norm": 1.5997624397277832,
"learning_rate": 9.625056467399489e-06,
"loss": 0.1483,
"step": 350
},
{
"epoch": 0.17805475183618963,
"grad_norm": 1.787446141242981,
"learning_rate": 9.54976660141545e-06,
"loss": 0.1493,
"step": 400
},
{
"epoch": 0.20031159581571334,
"grad_norm": 1.745363473892212,
"learning_rate": 9.474476735431411e-06,
"loss": 0.1518,
"step": 450
},
{
"epoch": 0.22256843979523702,
"grad_norm": 1.36530601978302,
"learning_rate": 9.399186869447373e-06,
"loss": 0.1378,
"step": 500
},
{
"epoch": 0.24482528377476073,
"grad_norm": 1.6621060371398926,
"learning_rate": 9.323897003463336e-06,
"loss": 0.1433,
"step": 550
},
{
"epoch": 0.2670821277542844,
"grad_norm": 1.249980092048645,
"learning_rate": 9.248607137479296e-06,
"loss": 0.1405,
"step": 600
},
{
"epoch": 0.2893389717338081,
"grad_norm": 1.4200546741485596,
"learning_rate": 9.173317271495258e-06,
"loss": 0.146,