tigre-llm-gemma4 / trainer_state.json
beshiribrahim's picture
checkpoint-57000 from tigre training
5f3048c verified
Raw
History Blame Contribute Delete
98.6 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9294892701063205,
"eval_steps": 500,
"global_step": 57000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0016306829300110886,
"grad_norm": 1848.0,
"learning_rate": 2.4750000000000002e-05,
"loss": 57.631083984375,
"step": 100
},
{
"epoch": 0.003261365860022177,
"grad_norm": 137.0,
"learning_rate": 4.975e-05,
"loss": 51.209560546875,
"step": 200
},
{
"epoch": 0.004892048790033266,
"grad_norm": 124.5,
"learning_rate": 4.9999676354266065e-05,
"loss": 49.47775390625,
"step": 300
},
{
"epoch": 0.006522731720044354,
"grad_norm": 79.5,
"learning_rate": 4.999869231602613e-05,
"loss": 48.704814453125,
"step": 400
},
{
"epoch": 0.008153414650055443,
"grad_norm": 68.5,
"learning_rate": 4.999704787825411e-05,
"loss": 47.960693359375,
"step": 500
},
{
"epoch": 0.009784097580066532,
"grad_norm": 189.0,
"learning_rate": 4.9994743084391704e-05,
"loss": 47.783095703125,
"step": 600
},
{
"epoch": 0.01141478051007762,
"grad_norm": 72.0,
"learning_rate": 4.99917779953255e-05,
"loss": 47.388173828125,
"step": 700
},
{
"epoch": 0.013045463440088709,
"grad_norm": 63.0,
"learning_rate": 4.998815268938533e-05,
"loss": 45.9780810546875,
"step": 800
},
{
"epoch": 0.014676146370099797,
"grad_norm": 100.5,
"learning_rate": 4.998386726234221e-05,
"loss": 45.814921875,
"step": 900
},
{
"epoch": 0.016306829300110886,
"grad_norm": 59.5,
"learning_rate": 4.997892182740583e-05,
"loss": 45.6537255859375,
"step": 1000
},
{
"epoch": 0.017937512230121976,
"grad_norm": 65.0,
"learning_rate": 4.997331651522155e-05,
"loss": 45.7083984375,
"step": 1100
},
{
"epoch": 0.019568195160133063,
"grad_norm": 75.5,
"learning_rate": 4.996705147386691e-05,
"loss": 44.537646484375,
"step": 1200
},
{
"epoch": 0.021198878090144153,
"grad_norm": 68.0,
"learning_rate": 4.996012686884781e-05,
"loss": 44.7698046875,
"step": 1300
},
{
"epoch": 0.02282956102015524,
"grad_norm": 69.5,
"learning_rate": 4.9952542883094035e-05,
"loss": 44.1380029296875,
"step": 1400
},
{
"epoch": 0.02446024395016633,
"grad_norm": 66.5,
"learning_rate": 4.9944299716954504e-05,
"loss": 43.58759765625,
"step": 1500
},
{
"epoch": 0.026090926880177417,
"grad_norm": 122.0,
"learning_rate": 4.9935397588191925e-05,
"loss": 44.7011474609375,
"step": 1600
},
{
"epoch": 0.027721609810188508,
"grad_norm": 109.5,
"learning_rate": 4.992583673197707e-05,
"loss": 44.5887255859375,
"step": 1700
},
{
"epoch": 0.029352292740199595,
"grad_norm": 68.5,
"learning_rate": 4.9915617400882534e-05,
"loss": 43.7655419921875,
"step": 1800
},
{
"epoch": 0.030982975670210685,
"grad_norm": 71.5,
"learning_rate": 4.990473986487611e-05,
"loss": 44.11833984375,
"step": 1900
},
{
"epoch": 0.03261365860022177,
"grad_norm": 52.0,
"learning_rate": 4.989320441131363e-05,
"loss": 42.64595703125,
"step": 2000
},
{
"epoch": 0.03424434153023286,
"grad_norm": 87.5,
"learning_rate": 4.9881011344931334e-05,
"loss": 42.78341796875,
"step": 2100
},
{
"epoch": 0.03587502446024395,
"grad_norm": 56.25,
"learning_rate": 4.98681609878379e-05,
"loss": 41.2956884765625,
"step": 2200
},
{
"epoch": 0.037505707390255036,
"grad_norm": 56.25,
"learning_rate": 4.9854653679505856e-05,
"loss": 43.3965185546875,
"step": 2300
},
{
"epoch": 0.039136390320266126,
"grad_norm": 68.5,
"learning_rate": 4.98404897767627e-05,
"loss": 42.278310546875,
"step": 2400
},
{
"epoch": 0.040767073250277217,
"grad_norm": 55.25,
"learning_rate": 4.982566965378138e-05,
"loss": 42.1446875,
"step": 2500
},
{
"epoch": 0.04239775618028831,
"grad_norm": 64.0,
"learning_rate": 4.981019370207046e-05,
"loss": 42.3336328125,
"step": 2600
},
{
"epoch": 0.04402843911029939,
"grad_norm": 66.0,
"learning_rate": 4.9794062330463786e-05,
"loss": 41.4570849609375,
"step": 2700
},
{
"epoch": 0.04565912204031048,
"grad_norm": 70.0,
"learning_rate": 4.977727596510966e-05,
"loss": 41.9864208984375,
"step": 2800
},
{
"epoch": 0.04728980497032157,
"grad_norm": 83.5,
"learning_rate": 4.975983504945958e-05,
"loss": 41.5266943359375,
"step": 2900
},
{
"epoch": 0.04892048790033266,
"grad_norm": 56.25,
"learning_rate": 4.974174004425656e-05,
"loss": 41.5351513671875,
"step": 3000
},
{
"epoch": 0.050551170830343745,
"grad_norm": 47.75,
"learning_rate": 4.9722991427522906e-05,
"loss": 41.425693359375,
"step": 3100
},
{
"epoch": 0.052181853760354835,
"grad_norm": 78.5,
"learning_rate": 4.970358969454765e-05,
"loss": 41.6533154296875,
"step": 3200
},
{
"epoch": 0.053812536690365925,
"grad_norm": 76.0,
"learning_rate": 4.9683535357873414e-05,
"loss": 41.316328125,
"step": 3300
},
{
"epoch": 0.055443219620377016,
"grad_norm": 63.75,
"learning_rate": 4.966282894728288e-05,
"loss": 40.73615234375,
"step": 3400
},
{
"epoch": 0.057073902550388106,
"grad_norm": 57.0,
"learning_rate": 4.964147100978483e-05,
"loss": 40.691181640625,
"step": 3500
},
{
"epoch": 0.05870458548039919,
"grad_norm": 56.0,
"learning_rate": 4.961946210959965e-05,
"loss": 40.45117919921875,
"step": 3600
},
{
"epoch": 0.06033526841041028,
"grad_norm": 58.75,
"learning_rate": 4.959680282814445e-05,
"loss": 40.51891845703125,
"step": 3700
},
{
"epoch": 0.06196595134042137,
"grad_norm": 66.0,
"learning_rate": 4.9573493764017684e-05,
"loss": 40.84674072265625,
"step": 3800
},
{
"epoch": 0.06359663427043245,
"grad_norm": 48.0,
"learning_rate": 4.954953553298337e-05,
"loss": 40.84188232421875,
"step": 3900
},
{
"epoch": 0.06522731720044354,
"grad_norm": 74.5,
"learning_rate": 4.9524928767954806e-05,
"loss": 40.9570556640625,
"step": 4000
},
{
"epoch": 0.06685800013045463,
"grad_norm": 68.0,
"learning_rate": 4.9499674118977824e-05,
"loss": 40.75696533203125,
"step": 4100
},
{
"epoch": 0.06848868306046572,
"grad_norm": 54.5,
"learning_rate": 4.947377225321366e-05,
"loss": 40.3128271484375,
"step": 4200
},
{
"epoch": 0.07011936599047681,
"grad_norm": 56.0,
"learning_rate": 4.944722385492132e-05,
"loss": 40.23645263671875,
"step": 4300
},
{
"epoch": 0.0717500489204879,
"grad_norm": 70.5,
"learning_rate": 4.9420029625439456e-05,
"loss": 40.52634521484375,
"step": 4400
},
{
"epoch": 0.073380731850499,
"grad_norm": 69.0,
"learning_rate": 4.939219028316793e-05,
"loss": 41.236279296875,
"step": 4500
},
{
"epoch": 0.07501141478051007,
"grad_norm": 58.0,
"learning_rate": 4.936370656354873e-05,
"loss": 39.47917724609375,
"step": 4600
},
{
"epoch": 0.07664209771052116,
"grad_norm": 55.0,
"learning_rate": 4.9334579219046646e-05,
"loss": 39.72600830078125,
"step": 4700
},
{
"epoch": 0.07827278064053225,
"grad_norm": 95.0,
"learning_rate": 4.930480901912929e-05,
"loss": 39.95852783203125,
"step": 4800
},
{
"epoch": 0.07990346357054334,
"grad_norm": 78.5,
"learning_rate": 4.927439675024686e-05,
"loss": 40.44654541015625,
"step": 4900
},
{
"epoch": 0.08153414650055443,
"grad_norm": 96.0,
"learning_rate": 4.92433432158113e-05,
"loss": 39.4082470703125,
"step": 5000
},
{
"epoch": 0.08316482943056552,
"grad_norm": 61.25,
"learning_rate": 4.92116492361751e-05,
"loss": 40.31276123046875,
"step": 5100
},
{
"epoch": 0.08479551236057661,
"grad_norm": 52.25,
"learning_rate": 4.917931564860964e-05,
"loss": 39.2943212890625,
"step": 5200
},
{
"epoch": 0.0864261952905877,
"grad_norm": 67.5,
"learning_rate": 4.9146343307283026e-05,
"loss": 39.6194384765625,
"step": 5300
},
{
"epoch": 0.08805687822059878,
"grad_norm": 54.0,
"learning_rate": 4.911273308323758e-05,
"loss": 39.684970703125,
"step": 5400
},
{
"epoch": 0.08968756115060987,
"grad_norm": 51.25,
"learning_rate": 4.9078485864366806e-05,
"loss": 39.5464306640625,
"step": 5500
},
{
"epoch": 0.09131824408062096,
"grad_norm": 58.25,
"learning_rate": 4.9043602555391905e-05,
"loss": 39.32220947265625,
"step": 5600
},
{
"epoch": 0.09294892701063205,
"grad_norm": 74.0,
"learning_rate": 4.900808407783792e-05,
"loss": 39.3750537109375,
"step": 5700
},
{
"epoch": 0.09457960994064314,
"grad_norm": 59.0,
"learning_rate": 4.897193137000938e-05,
"loss": 39.05360595703125,
"step": 5800
},
{
"epoch": 0.09621029287065423,
"grad_norm": 68.0,
"learning_rate": 4.893514538696547e-05,
"loss": 39.27678955078125,
"step": 5900
},
{
"epoch": 0.09784097580066532,
"grad_norm": 69.0,
"learning_rate": 4.8897727100494894e-05,
"loss": 39.49569091796875,
"step": 6000
},
{
"epoch": 0.09947165873067641,
"grad_norm": 75.0,
"learning_rate": 4.885967749909009e-05,
"loss": 39.45048095703125,
"step": 6100
},
{
"epoch": 0.10110234166068749,
"grad_norm": 66.0,
"learning_rate": 4.88209975879212e-05,
"loss": 39.08032958984375,
"step": 6200
},
{
"epoch": 0.10273302459069858,
"grad_norm": 56.25,
"learning_rate": 4.878168838880948e-05,
"loss": 38.8167919921875,
"step": 6300
},
{
"epoch": 0.10436370752070967,
"grad_norm": 63.75,
"learning_rate": 4.874175094020033e-05,
"loss": 38.8155615234375,
"step": 6400
},
{
"epoch": 0.10599439045072076,
"grad_norm": 66.0,
"learning_rate": 4.8701186297135806e-05,
"loss": 38.7705859375,
"step": 6500
},
{
"epoch": 0.10762507338073185,
"grad_norm": 68.0,
"learning_rate": 4.8659995531226835e-05,
"loss": 38.30322021484375,
"step": 6600
},
{
"epoch": 0.10925575631074294,
"grad_norm": 57.75,
"learning_rate": 4.861817973062483e-05,
"loss": 38.99251220703125,
"step": 6700
},
{
"epoch": 0.11088643924075403,
"grad_norm": 57.75,
"learning_rate": 4.857573999999297e-05,
"loss": 37.8694775390625,
"step": 6800
},
{
"epoch": 0.11251712217076512,
"grad_norm": 59.75,
"learning_rate": 4.853267746047704e-05,
"loss": 38.29063720703125,
"step": 6900
},
{
"epoch": 0.11414780510077621,
"grad_norm": 55.25,
"learning_rate": 4.8488993249675786e-05,
"loss": 38.3592333984375,
"step": 7000
},
{
"epoch": 0.11577848803078729,
"grad_norm": 66.0,
"learning_rate": 4.844468852161086e-05,
"loss": 37.94030029296875,
"step": 7100
},
{
"epoch": 0.11740917096079838,
"grad_norm": 56.0,
"learning_rate": 4.839976444669634e-05,
"loss": 38.8104248046875,
"step": 7200
},
{
"epoch": 0.11903985389080947,
"grad_norm": 64.5,
"learning_rate": 4.835422221170785e-05,
"loss": 38.6752197265625,
"step": 7300
},
{
"epoch": 0.12067053682082056,
"grad_norm": 66.0,
"learning_rate": 4.8308063019751126e-05,
"loss": 38.39019287109375,
"step": 7400
},
{
"epoch": 0.12230121975083165,
"grad_norm": 57.75,
"learning_rate": 4.826128809023032e-05,
"loss": 38.142490234375,
"step": 7500
},
{
"epoch": 0.12393190268084274,
"grad_norm": 73.0,
"learning_rate": 4.821389865881571e-05,
"loss": 38.713876953125,
"step": 7600
},
{
"epoch": 0.12556258561085382,
"grad_norm": 81.5,
"learning_rate": 4.816589597741114e-05,
"loss": 38.9951513671875,
"step": 7700
},
{
"epoch": 0.1271932685408649,
"grad_norm": 69.5,
"learning_rate": 4.811728131412085e-05,
"loss": 38.17974609375,
"step": 7800
},
{
"epoch": 0.128823951470876,
"grad_norm": 63.0,
"learning_rate": 4.806805595321606e-05,
"loss": 38.03181396484375,
"step": 7900
},
{
"epoch": 0.1304546344008871,
"grad_norm": 75.5,
"learning_rate": 4.8018221195101007e-05,
"loss": 38.01936279296875,
"step": 8000
},
{
"epoch": 0.13208531733089818,
"grad_norm": 52.75,
"learning_rate": 4.796777835627858e-05,
"loss": 38.16745849609375,
"step": 8100
},
{
"epoch": 0.13371600026090927,
"grad_norm": 67.0,
"learning_rate": 4.7916728769315585e-05,
"loss": 38.12496826171875,
"step": 8200
},
{
"epoch": 0.13534668319092036,
"grad_norm": 54.5,
"learning_rate": 4.786507378280749e-05,
"loss": 37.94025634765625,
"step": 8300
},
{
"epoch": 0.13697736612093145,
"grad_norm": 57.0,
"learning_rate": 4.781281476134283e-05,
"loss": 37.59670654296875,
"step": 8400
},
{
"epoch": 0.13860804905094254,
"grad_norm": 81.0,
"learning_rate": 4.7759953085467144e-05,
"loss": 38.01157958984375,
"step": 8500
},
{
"epoch": 0.14023873198095363,
"grad_norm": 64.0,
"learning_rate": 4.7706490151646525e-05,
"loss": 38.13415771484375,
"step": 8600
},
{
"epoch": 0.14186941491096472,
"grad_norm": 56.5,
"learning_rate": 4.7652427372230714e-05,
"loss": 37.7507568359375,
"step": 8700
},
{
"epoch": 0.1435000978409758,
"grad_norm": 64.5,
"learning_rate": 4.759776617541577e-05,
"loss": 37.76613037109375,
"step": 8800
},
{
"epoch": 0.1451307807709869,
"grad_norm": 68.0,
"learning_rate": 4.754250800520639e-05,
"loss": 37.7645263671875,
"step": 8900
},
{
"epoch": 0.146761463700998,
"grad_norm": 63.75,
"learning_rate": 4.748665432137771e-05,
"loss": 37.4073974609375,
"step": 9000
},
{
"epoch": 0.14839214663100905,
"grad_norm": 60.5,
"learning_rate": 4.743020659943679e-05,
"loss": 37.92957763671875,
"step": 9100
},
{
"epoch": 0.15002282956102014,
"grad_norm": 62.0,
"learning_rate": 4.7373166330583586e-05,
"loss": 37.5009765625,
"step": 9200
},
{
"epoch": 0.15165351249103123,
"grad_norm": 68.5,
"learning_rate": 4.7315535021671596e-05,
"loss": 36.90603759765625,
"step": 9300
},
{
"epoch": 0.15328419542104232,
"grad_norm": 58.0,
"learning_rate": 4.725731419516804e-05,
"loss": 37.43829345703125,
"step": 9400
},
{
"epoch": 0.15491487835105341,
"grad_norm": 73.0,
"learning_rate": 4.719850538911361e-05,
"loss": 37.32279296875,
"step": 9500
},
{
"epoch": 0.1565455612810645,
"grad_norm": 77.5,
"learning_rate": 4.7139110157081926e-05,
"loss": 37.51306396484375,
"step": 9600
},
{
"epoch": 0.1581762442110756,
"grad_norm": 64.0,
"learning_rate": 4.707913006813836e-05,
"loss": 37.788876953125,
"step": 9700
},
{
"epoch": 0.15980692714108669,
"grad_norm": 70.5,
"learning_rate": 4.7018566706798726e-05,
"loss": 37.37509765625,
"step": 9800
},
{
"epoch": 0.16143761007109778,
"grad_norm": 77.0,
"learning_rate": 4.695742167298733e-05,
"loss": 37.49742431640625,
"step": 9900
},
{
"epoch": 0.16306829300110887,
"grad_norm": 74.5,
"learning_rate": 4.689569658199472e-05,
"loss": 36.263544921875,
"step": 10000
},
{
"epoch": 0.16469897593111996,
"grad_norm": 54.5,
"learning_rate": 4.683339306443505e-05,
"loss": 37.2926611328125,
"step": 10100
},
{
"epoch": 0.16632965886113105,
"grad_norm": 62.75,
"learning_rate": 4.677051276620295e-05,
"loss": 36.614912109375,
"step": 10200
},
{
"epoch": 0.16796034179114214,
"grad_norm": 59.5,
"learning_rate": 4.6707057348430114e-05,
"loss": 38.0705517578125,
"step": 10300
},
{
"epoch": 0.16959102472115323,
"grad_norm": 59.75,
"learning_rate": 4.664302848744134e-05,
"loss": 37.4716015625,
"step": 10400
},
{
"epoch": 0.17122170765116432,
"grad_norm": 75.0,
"learning_rate": 4.657842787471029e-05,
"loss": 36.26463623046875,
"step": 10500
},
{
"epoch": 0.1728523905811754,
"grad_norm": 68.5,
"learning_rate": 4.651325721681483e-05,
"loss": 37.1750927734375,
"step": 10600
},
{
"epoch": 0.1744830735111865,
"grad_norm": 65.5,
"learning_rate": 4.6447518235391874e-05,
"loss": 37.45011474609375,
"step": 10700
},
{
"epoch": 0.17611375644119756,
"grad_norm": 63.75,
"learning_rate": 4.6381212667091974e-05,
"loss": 36.583759765625,
"step": 10800
},
{
"epoch": 0.17774443937120865,
"grad_norm": 81.5,
"learning_rate": 4.6314342263533396e-05,
"loss": 37.8414892578125,
"step": 10900
},
{
"epoch": 0.17937512230121974,
"grad_norm": 67.0,
"learning_rate": 4.6246908791255876e-05,
"loss": 36.96684326171875,
"step": 11000
},
{
"epoch": 0.18100580523123083,
"grad_norm": 65.5,
"learning_rate": 4.617891403167394e-05,
"loss": 36.708056640625,
"step": 11100
},
{
"epoch": 0.18263648816124192,
"grad_norm": 78.0,
"learning_rate": 4.6110359781029835e-05,
"loss": 37.07576904296875,
"step": 11200
},
{
"epoch": 0.184267171091253,
"grad_norm": 71.5,
"learning_rate": 4.6041247850346104e-05,
"loss": 37.048505859375,
"step": 11300
},
{
"epoch": 0.1858978540212641,
"grad_norm": 63.5,
"learning_rate": 4.59715800653777e-05,
"loss": 37.1758349609375,
"step": 11400
},
{
"epoch": 0.1875285369512752,
"grad_norm": 82.5,
"learning_rate": 4.59013582665638e-05,
"loss": 36.46238037109375,
"step": 11500
},
{
"epoch": 0.18915921988128628,
"grad_norm": 61.0,
"learning_rate": 4.583058430897917e-05,
"loss": 36.9132421875,
"step": 11600
},
{
"epoch": 0.19078990281129737,
"grad_norm": 60.5,
"learning_rate": 4.5759260062285136e-05,
"loss": 36.6371044921875,
"step": 11700
},
{
"epoch": 0.19242058574130846,
"grad_norm": 62.75,
"learning_rate": 4.568738741068023e-05,
"loss": 37.21307373046875,
"step": 11800
},
{
"epoch": 0.19405126867131955,
"grad_norm": 79.5,
"learning_rate": 4.5614968252850375e-05,
"loss": 36.160966796875,
"step": 11900
},
{
"epoch": 0.19568195160133064,
"grad_norm": 61.5,
"learning_rate": 4.554200450191877e-05,
"loss": 36.407998046875,
"step": 12000
},
{
"epoch": 0.19731263453134174,
"grad_norm": 86.0,
"learning_rate": 4.546849808539531e-05,
"loss": 36.465556640625,
"step": 12100
},
{
"epoch": 0.19894331746135283,
"grad_norm": 82.0,
"learning_rate": 4.5394450945125695e-05,
"loss": 36.30413330078125,
"step": 12200
},
{
"epoch": 0.20057400039136392,
"grad_norm": 74.5,
"learning_rate": 4.531986503724012e-05,
"loss": 37.130224609375,
"step": 12300
},
{
"epoch": 0.20220468332137498,
"grad_norm": 58.0,
"learning_rate": 4.524474233210158e-05,
"loss": 37.06889404296875,
"step": 12400
},
{
"epoch": 0.20383536625138607,
"grad_norm": 66.5,
"learning_rate": 4.516908481425387e-05,
"loss": 35.63563232421875,
"step": 12500
},
{
"epoch": 0.20546604918139716,
"grad_norm": 75.0,
"learning_rate": 4.509289448236911e-05,
"loss": 36.49399658203125,
"step": 12600
},
{
"epoch": 0.20709673211140825,
"grad_norm": 71.5,
"learning_rate": 4.5016173349194954e-05,
"loss": 37.03541748046875,
"step": 12700
},
{
"epoch": 0.20872741504141934,
"grad_norm": 70.5,
"learning_rate": 4.493892344150144e-05,
"loss": 37.067001953125,
"step": 12800
},
{
"epoch": 0.21035809797143043,
"grad_norm": 67.0,
"learning_rate": 4.486114680002743e-05,
"loss": 36.4354638671875,
"step": 12900
},
{
"epoch": 0.21198878090144152,
"grad_norm": 73.0,
"learning_rate": 4.47828454794267e-05,
"loss": 36.40055419921875,
"step": 13000
},
{
"epoch": 0.2136194638314526,
"grad_norm": 78.0,
"learning_rate": 4.470402154821367e-05,
"loss": 36.3733251953125,
"step": 13100
},
{
"epoch": 0.2152501467614637,
"grad_norm": 76.0,
"learning_rate": 4.4624677088708755e-05,
"loss": 35.84779296875,
"step": 13200
},
{
"epoch": 0.2168808296914748,
"grad_norm": 72.0,
"learning_rate": 4.4544814196983345e-05,
"loss": 36.6591943359375,
"step": 13300
},
{
"epoch": 0.21851151262148588,
"grad_norm": 70.5,
"learning_rate": 4.446443498280446e-05,
"loss": 36.5290478515625,
"step": 13400
},
{
"epoch": 0.22014219555149697,
"grad_norm": 72.5,
"learning_rate": 4.438354156957898e-05,
"loss": 36.95695556640625,
"step": 13500
},
{
"epoch": 0.22177287848150806,
"grad_norm": 82.0,
"learning_rate": 4.430213609429757e-05,
"loss": 37.469677734375,
"step": 13600
},
{
"epoch": 0.22340356141151915,
"grad_norm": 64.5,
"learning_rate": 4.422022070747825e-05,
"loss": 36.18645751953125,
"step": 13700
},
{
"epoch": 0.22503424434153024,
"grad_norm": 93.5,
"learning_rate": 4.413779757310951e-05,
"loss": 36.1962841796875,
"step": 13800
},
{
"epoch": 0.22666492727154133,
"grad_norm": 71.5,
"learning_rate": 4.405486886859323e-05,
"loss": 36.54798095703125,
"step": 13900
},
{
"epoch": 0.22829561020155242,
"grad_norm": 71.5,
"learning_rate": 4.397143678468712e-05,
"loss": 37.03234130859375,
"step": 14000
},
{
"epoch": 0.2299262931315635,
"grad_norm": 68.5,
"learning_rate": 4.388750352544682e-05,
"loss": 36.26530517578125,
"step": 14100
},
{
"epoch": 0.23155697606157458,
"grad_norm": 69.0,
"learning_rate": 4.380307130816772e-05,
"loss": 35.5451171875,
"step": 14200
},
{
"epoch": 0.23318765899158567,
"grad_norm": 68.0,
"learning_rate": 4.371814236332637e-05,
"loss": 36.003154296875,
"step": 14300
},
{
"epoch": 0.23481834192159676,
"grad_norm": 81.5,
"learning_rate": 4.363271893452153e-05,
"loss": 36.36017578125,
"step": 14400
},
{
"epoch": 0.23644902485160785,
"grad_norm": 72.0,
"learning_rate": 4.354680327841496e-05,
"loss": 36.13802978515625,
"step": 14500
},
{
"epoch": 0.23807970778161894,
"grad_norm": 84.5,
"learning_rate": 4.3460397664671714e-05,
"loss": 35.47357666015625,
"step": 14600
},
{
"epoch": 0.23971039071163003,
"grad_norm": 76.0,
"learning_rate": 4.3373504375900284e-05,
"loss": 35.66486328125,
"step": 14700
},
{
"epoch": 0.24134107364164112,
"grad_norm": 62.25,
"learning_rate": 4.328612570759223e-05,
"loss": 36.26236083984375,
"step": 14800
},
{
"epoch": 0.2429717565716522,
"grad_norm": 68.5,
"learning_rate": 4.3198263968061556e-05,
"loss": 36.83368408203125,
"step": 14900
},
{
"epoch": 0.2446024395016633,
"grad_norm": 71.0,
"learning_rate": 4.310992147838373e-05,
"loss": 36.66268798828125,
"step": 15000
},
{
"epoch": 0.2462331224316744,
"grad_norm": 74.0,
"learning_rate": 4.3021100572334396e-05,
"loss": 36.68743896484375,
"step": 15100
},
{
"epoch": 0.24786380536168548,
"grad_norm": 71.5,
"learning_rate": 4.293180359632767e-05,
"loss": 35.35719970703125,
"step": 15200
},
{
"epoch": 0.24949448829169657,
"grad_norm": 67.0,
"learning_rate": 4.28420329093542e-05,
"loss": 36.01685302734375,
"step": 15300
},
{
"epoch": 0.25112517122170763,
"grad_norm": 75.5,
"learning_rate": 4.2751790882918805e-05,
"loss": 36.48025146484375,
"step": 15400
},
{
"epoch": 0.25275585415171875,
"grad_norm": 75.0,
"learning_rate": 4.266107990097788e-05,
"loss": 36.5563720703125,
"step": 15500
},
{
"epoch": 0.2543865370817298,
"grad_norm": 75.0,
"learning_rate": 4.256990235987636e-05,
"loss": 35.87431884765625,
"step": 15600
},
{
"epoch": 0.25601722001174093,
"grad_norm": 80.0,
"learning_rate": 4.247826066828447e-05,
"loss": 35.52444091796875,
"step": 15700
},
{
"epoch": 0.257647902941752,
"grad_norm": 72.5,
"learning_rate": 4.238615724713404e-05,
"loss": 35.4122216796875,
"step": 15800
},
{
"epoch": 0.2592785858717631,
"grad_norm": 63.25,
"learning_rate": 4.22935945295546e-05,
"loss": 35.93663330078125,
"step": 15900
},
{
"epoch": 0.2609092688017742,
"grad_norm": 65.5,
"learning_rate": 4.220057496080906e-05,
"loss": 35.59571044921875,
"step": 16000
},
{
"epoch": 0.2625399517317853,
"grad_norm": 80.0,
"learning_rate": 4.210710099822915e-05,
"loss": 35.44169677734375,
"step": 16100
},
{
"epoch": 0.26417063466179636,
"grad_norm": 67.5,
"learning_rate": 4.201317511115047e-05,
"loss": 35.58749755859375,
"step": 16200
},
{
"epoch": 0.2658013175918075,
"grad_norm": 68.5,
"learning_rate": 4.1918799780847304e-05,
"loss": 36.0799951171875,
"step": 16300
},
{
"epoch": 0.26743200052181854,
"grad_norm": 65.0,
"learning_rate": 4.1823977500467006e-05,
"loss": 35.81424560546875,
"step": 16400
},
{
"epoch": 0.26906268345182965,
"grad_norm": 68.0,
"learning_rate": 4.1728710774964186e-05,
"loss": 35.78817626953125,
"step": 16500
},
{
"epoch": 0.2706933663818407,
"grad_norm": 75.5,
"learning_rate": 4.1633002121034535e-05,
"loss": 35.50345458984375,
"step": 16600
},
{
"epoch": 0.2723240493118518,
"grad_norm": 74.5,
"learning_rate": 4.1536854067048315e-05,
"loss": 35.90816650390625,
"step": 16700
},
{
"epoch": 0.2739547322418629,
"grad_norm": 63.0,
"learning_rate": 4.144026915298357e-05,
"loss": 35.30243408203125,
"step": 16800
},
{
"epoch": 0.27558541517187396,
"grad_norm": 74.5,
"learning_rate": 4.134324993035904e-05,
"loss": 35.09135986328125,
"step": 16900
},
{
"epoch": 0.2772160981018851,
"grad_norm": 71.5,
"learning_rate": 4.124579896216677e-05,
"loss": 35.6159228515625,
"step": 17000
},
{
"epoch": 0.27884678103189614,
"grad_norm": 79.0,
"learning_rate": 4.1147918822804385e-05,
"loss": 35.941611328125,
"step": 17100
},
{
"epoch": 0.28047746396190726,
"grad_norm": 65.5,
"learning_rate": 4.104961209800704e-05,
"loss": 36.71083740234375,
"step": 17200
},
{
"epoch": 0.2821081468919183,
"grad_norm": 94.0,
"learning_rate": 4.0950881384779186e-05,
"loss": 35.25750244140625,
"step": 17300
},
{
"epoch": 0.28373882982192944,
"grad_norm": 91.5,
"learning_rate": 4.0851729291325944e-05,
"loss": 35.74835693359375,
"step": 17400
},
{
"epoch": 0.2853695127519405,
"grad_norm": 103.0,
"learning_rate": 4.0752158436984146e-05,
"loss": 35.55155029296875,
"step": 17500
},
{
"epoch": 0.2870001956819516,
"grad_norm": 67.0,
"learning_rate": 4.065217145215321e-05,
"loss": 35.276826171875,
"step": 17600
},
{
"epoch": 0.2886308786119627,
"grad_norm": 75.0,
"learning_rate": 4.055177097822564e-05,
"loss": 35.8741796875,
"step": 17700
},
{
"epoch": 0.2902615615419738,
"grad_norm": 74.0,
"learning_rate": 4.045095966751718e-05,
"loss": 34.55402587890625,
"step": 17800
},
{
"epoch": 0.29189224447198486,
"grad_norm": 63.25,
"learning_rate": 4.0349740183196855e-05,
"loss": 36.291943359375,
"step": 17900
},
{
"epoch": 0.293522927401996,
"grad_norm": 73.0,
"learning_rate": 4.024811519921653e-05,
"loss": 34.70413818359375,
"step": 18000
},
{
"epoch": 0.29515361033200704,
"grad_norm": 90.0,
"learning_rate": 4.0146087400240296e-05,
"loss": 35.33369384765625,
"step": 18100
},
{
"epoch": 0.2967842932620181,
"grad_norm": 65.0,
"learning_rate": 4.004365948157358e-05,
"loss": 35.17259033203125,
"step": 18200
},
{
"epoch": 0.2984149761920292,
"grad_norm": 76.0,
"learning_rate": 3.99408341490919e-05,
"loss": 35.45167724609375,
"step": 18300
},
{
"epoch": 0.3000456591220403,
"grad_norm": 67.5,
"learning_rate": 3.9837614119169395e-05,
"loss": 36.00077880859375,
"step": 18400
},
{
"epoch": 0.3016763420520514,
"grad_norm": 75.0,
"learning_rate": 3.973400211860707e-05,
"loss": 35.71475341796875,
"step": 18500
},
{
"epoch": 0.30330702498206247,
"grad_norm": 74.5,
"learning_rate": 3.963000088456078e-05,
"loss": 35.9516943359375,
"step": 18600
},
{
"epoch": 0.3049377079120736,
"grad_norm": 70.5,
"learning_rate": 3.9525613164468884e-05,
"loss": 35.4289453125,
"step": 18700
},
{
"epoch": 0.30656839084208465,
"grad_norm": 86.5,
"learning_rate": 3.9420841715979684e-05,
"loss": 34.90598876953125,
"step": 18800
},
{
"epoch": 0.30819907377209577,
"grad_norm": 75.5,
"learning_rate": 3.931568930687857e-05,
"loss": 35.905595703125,
"step": 18900
},
{
"epoch": 0.30982975670210683,
"grad_norm": 91.0,
"learning_rate": 3.921015871501493e-05,
"loss": 34.45498779296875,
"step": 19000
},
{
"epoch": 0.31146043963211795,
"grad_norm": 80.0,
"learning_rate": 3.910425272822872e-05,
"loss": 34.62409423828125,
"step": 19100
},
{
"epoch": 0.313091122562129,
"grad_norm": 63.75,
"learning_rate": 3.899797414427685e-05,
"loss": 35.5582763671875,
"step": 19200
},
{
"epoch": 0.3147218054921401,
"grad_norm": 75.0,
"learning_rate": 3.889132577075927e-05,
"loss": 34.56370849609375,
"step": 19300
},
{
"epoch": 0.3163524884221512,
"grad_norm": 76.5,
"learning_rate": 3.8784310425044766e-05,
"loss": 35.09487548828125,
"step": 19400
},
{
"epoch": 0.3179831713521623,
"grad_norm": 65.0,
"learning_rate": 3.8676930934196606e-05,
"loss": 35.8907958984375,
"step": 19500
},
{
"epoch": 0.31961385428217337,
"grad_norm": 69.5,
"learning_rate": 3.856919013489778e-05,
"loss": 35.13558837890625,
"step": 19600
},
{
"epoch": 0.3212445372121845,
"grad_norm": 64.0,
"learning_rate": 3.84610908733761e-05,
"loss": 35.35878662109375,
"step": 19700
},
{
"epoch": 0.32287522014219555,
"grad_norm": 66.5,
"learning_rate": 3.8352636005329024e-05,
"loss": 35.32779052734375,
"step": 19800
},
{
"epoch": 0.3245059030722066,
"grad_norm": 87.0,
"learning_rate": 3.824382839584817e-05,
"loss": 35.43947021484375,
"step": 19900
},
{
"epoch": 0.32613658600221773,
"grad_norm": 72.5,
"learning_rate": 3.8134670919343666e-05,
"loss": 35.74707275390625,
"step": 20000
},
{
"epoch": 0.3277672689322288,
"grad_norm": 73.5,
"learning_rate": 3.802516645946822e-05,
"loss": 35.45427734375,
"step": 20100
},
{
"epoch": 0.3293979518622399,
"grad_norm": 85.5,
"learning_rate": 3.791531790904088e-05,
"loss": 35.19885498046875,
"step": 20200
},
{
"epoch": 0.331028634792251,
"grad_norm": 73.0,
"learning_rate": 3.780512816997072e-05,
"loss": 34.578623046875,
"step": 20300
},
{
"epoch": 0.3326593177222621,
"grad_norm": 62.5,
"learning_rate": 3.769460015318007e-05,
"loss": 35.428427734375,
"step": 20400
},
{
"epoch": 0.33429000065227316,
"grad_norm": 85.5,
"learning_rate": 3.758373677852769e-05,
"loss": 35.886455078125,
"step": 20500
},
{
"epoch": 0.3359206835822843,
"grad_norm": 66.5,
"learning_rate": 3.7472540974731594e-05,
"loss": 35.593017578125,
"step": 20600
},
{
"epoch": 0.33755136651229534,
"grad_norm": 82.5,
"learning_rate": 3.736101567929173e-05,
"loss": 34.89272216796875,
"step": 20700
},
{
"epoch": 0.33918204944230645,
"grad_norm": 63.5,
"learning_rate": 3.724916383841229e-05,
"loss": 35.02741943359375,
"step": 20800
},
{
"epoch": 0.3408127323723175,
"grad_norm": 64.5,
"learning_rate": 3.7136988406923994e-05,
"loss": 36.2602587890625,
"step": 20900
},
{
"epoch": 0.34244341530232864,
"grad_norm": 73.5,
"learning_rate": 3.702449234820594e-05,
"loss": 35.3712841796875,
"step": 21000
},
{
"epoch": 0.3440740982323397,
"grad_norm": 73.0,
"learning_rate": 3.691167863410736e-05,
"loss": 34.90469482421875,
"step": 21100
},
{
"epoch": 0.3457047811623508,
"grad_norm": 63.5,
"learning_rate": 3.679855024486909e-05,
"loss": 34.036279296875,
"step": 21200
},
{
"epoch": 0.3473354640923619,
"grad_norm": 72.0,
"learning_rate": 3.668511016904489e-05,
"loss": 35.082548828125,
"step": 21300
},
{
"epoch": 0.348966147022373,
"grad_norm": 67.0,
"learning_rate": 3.657136140342241e-05,
"loss": 35.16548095703125,
"step": 21400
},
{
"epoch": 0.35059682995238406,
"grad_norm": 72.0,
"learning_rate": 3.645730695294411e-05,
"loss": 34.25626220703125,
"step": 21500
},
{
"epoch": 0.3522275128823951,
"grad_norm": 58.75,
"learning_rate": 3.634294983062782e-05,
"loss": 35.02423095703125,
"step": 21600
},
{
"epoch": 0.35385819581240624,
"grad_norm": 87.0,
"learning_rate": 3.622829305748715e-05,
"loss": 34.51501708984375,
"step": 21700
},
{
"epoch": 0.3554888787424173,
"grad_norm": 66.5,
"learning_rate": 3.611333966245173e-05,
"loss": 34.61720703125,
"step": 21800
},
{
"epoch": 0.3571195616724284,
"grad_norm": 86.0,
"learning_rate": 3.599809268228712e-05,
"loss": 34.3772216796875,
"step": 21900
},
{
"epoch": 0.3587502446024395,
"grad_norm": 99.0,
"learning_rate": 3.588255516151465e-05,
"loss": 35.2338037109375,
"step": 22000
},
{
"epoch": 0.3603809275324506,
"grad_norm": 76.5,
"learning_rate": 3.576673015233096e-05,
"loss": 35.32531494140625,
"step": 22100
},
{
"epoch": 0.36201161046246166,
"grad_norm": 118.0,
"learning_rate": 3.565062071452737e-05,
"loss": 35.19863037109375,
"step": 22200
},
{
"epoch": 0.3636422933924728,
"grad_norm": 75.5,
"learning_rate": 3.553422991540906e-05,
"loss": 35.305537109375,
"step": 22300
},
{
"epoch": 0.36527297632248384,
"grad_norm": 100.5,
"learning_rate": 3.541756082971404e-05,
"loss": 34.02279296875,
"step": 22400
},
{
"epoch": 0.36690365925249496,
"grad_norm": 96.5,
"learning_rate": 3.530061653953189e-05,
"loss": 35.02393798828125,
"step": 22500
},
{
"epoch": 0.368534342182506,
"grad_norm": 67.5,
"learning_rate": 3.518340013422241e-05,
"loss": 34.391298828125,
"step": 22600
},
{
"epoch": 0.37016502511251714,
"grad_norm": 68.5,
"learning_rate": 3.506591471033392e-05,
"loss": 34.5220068359375,
"step": 22700
},
{
"epoch": 0.3717957080425282,
"grad_norm": 81.0,
"learning_rate": 3.494816337152154e-05,
"loss": 34.79232666015625,
"step": 22800
},
{
"epoch": 0.3734263909725393,
"grad_norm": 78.0,
"learning_rate": 3.483014922846514e-05,
"loss": 34.74611083984375,
"step": 22900
},
{
"epoch": 0.3750570739025504,
"grad_norm": 80.0,
"learning_rate": 3.471187539878718e-05,
"loss": 35.257373046875,
"step": 23000
},
{
"epoch": 0.3766877568325615,
"grad_norm": 81.0,
"learning_rate": 3.4593345006970366e-05,
"loss": 35.04670166015625,
"step": 23100
},
{
"epoch": 0.37831843976257257,
"grad_norm": 72.5,
"learning_rate": 3.447456118427511e-05,
"loss": 34.53382080078125,
"step": 23200
},
{
"epoch": 0.37994912269258363,
"grad_norm": 75.5,
"learning_rate": 3.435552706865677e-05,
"loss": 34.72044189453125,
"step": 23300
},
{
"epoch": 0.38157980562259475,
"grad_norm": 88.5,
"learning_rate": 3.4236245804682824e-05,
"loss": 34.974482421875,
"step": 23400
},
{
"epoch": 0.3832104885526058,
"grad_norm": 88.5,
"learning_rate": 3.4116720543449705e-05,
"loss": 34.9816455078125,
"step": 23500
},
{
"epoch": 0.38484117148261693,
"grad_norm": 106.5,
"learning_rate": 3.399695444249966e-05,
"loss": 34.2979931640625,
"step": 23600
},
{
"epoch": 0.386471854412628,
"grad_norm": 79.5,
"learning_rate": 3.387695066573726e-05,
"loss": 34.75494873046875,
"step": 23700
},
{
"epoch": 0.3881025373426391,
"grad_norm": 84.0,
"learning_rate": 3.375671238334584e-05,
"loss": 33.58097412109375,
"step": 23800
},
{
"epoch": 0.38973322027265017,
"grad_norm": 69.0,
"learning_rate": 3.363624277170378e-05,
"loss": 34.609541015625,
"step": 23900
},
{
"epoch": 0.3913639032026613,
"grad_norm": 78.0,
"learning_rate": 3.3515545013300556e-05,
"loss": 35.16354736328125,
"step": 24000
},
{
"epoch": 0.39299458613267235,
"grad_norm": 77.0,
"learning_rate": 3.3394622296652663e-05,
"loss": 35.11162353515625,
"step": 24100
},
{
"epoch": 0.39462526906268347,
"grad_norm": 80.5,
"learning_rate": 3.3273477816219416e-05,
"loss": 35.095224609375,
"step": 24200
},
{
"epoch": 0.39625595199269453,
"grad_norm": 94.0,
"learning_rate": 3.315211477231856e-05,
"loss": 34.56839111328125,
"step": 24300
},
{
"epoch": 0.39788663492270565,
"grad_norm": 75.5,
"learning_rate": 3.303053637104168e-05,
"loss": 35.119716796875,
"step": 24400
},
{
"epoch": 0.3995173178527167,
"grad_norm": 72.5,
"learning_rate": 3.290874582416954e-05,
"loss": 33.57674560546875,
"step": 24500
},
{
"epoch": 0.40114800078272783,
"grad_norm": 79.5,
"learning_rate": 3.278674634908726e-05,
"loss": 34.55984130859375,
"step": 24600
},
{
"epoch": 0.4027786837127389,
"grad_norm": 77.0,
"learning_rate": 3.266454116869926e-05,
"loss": 34.6054345703125,
"step": 24700
},
{
"epoch": 0.40440936664274996,
"grad_norm": 70.5,
"learning_rate": 3.254213351134416e-05,
"loss": 34.61232666015625,
"step": 24800
},
{
"epoch": 0.4060400495727611,
"grad_norm": 73.0,
"learning_rate": 3.24195266107095e-05,
"loss": 34.59481201171875,
"step": 24900
},
{
"epoch": 0.40767073250277214,
"grad_norm": 87.5,
"learning_rate": 3.229672370574629e-05,
"loss": 34.4932177734375,
"step": 25000
},
{
"epoch": 0.40930141543278326,
"grad_norm": 78.0,
"learning_rate": 3.217372804058346e-05,
"loss": 34.55974609375,
"step": 25100
},
{
"epoch": 0.4109320983627943,
"grad_norm": 73.0,
"learning_rate": 3.205054286444216e-05,
"loss": 35.383916015625,
"step": 25200
},
{
"epoch": 0.41256278129280544,
"grad_norm": 69.0,
"learning_rate": 3.1927171431549926e-05,
"loss": 34.20936767578125,
"step": 25300
},
{
"epoch": 0.4141934642228165,
"grad_norm": 85.0,
"learning_rate": 3.180361700105467e-05,
"loss": 34.470244140625,
"step": 25400
},
{
"epoch": 0.4158241471528276,
"grad_norm": 72.0,
"learning_rate": 3.167988283693867e-05,
"loss": 33.9093017578125,
"step": 25500
},
{
"epoch": 0.4174548300828387,
"grad_norm": 77.0,
"learning_rate": 3.155597220793226e-05,
"loss": 34.381337890625,
"step": 25600
},
{
"epoch": 0.4190855130128498,
"grad_norm": 85.0,
"learning_rate": 3.1431888387427524e-05,
"loss": 34.49350830078125,
"step": 25700
},
{
"epoch": 0.42071619594286086,
"grad_norm": 76.5,
"learning_rate": 3.1307634653391794e-05,
"loss": 34.64231201171875,
"step": 25800
},
{
"epoch": 0.422346878872872,
"grad_norm": 76.5,
"learning_rate": 3.118321428828109e-05,
"loss": 35.3302294921875,
"step": 25900
},
{
"epoch": 0.42397756180288304,
"grad_norm": 81.0,
"learning_rate": 3.1058630578953365e-05,
"loss": 33.86900634765625,
"step": 26000
},
{
"epoch": 0.42560824473289416,
"grad_norm": 64.5,
"learning_rate": 3.093388681658172e-05,
"loss": 34.8153857421875,
"step": 26100
},
{
"epoch": 0.4272389276629052,
"grad_norm": 80.5,
"learning_rate": 3.08089862965674e-05,
"loss": 34.62360595703125,
"step": 26200
},
{
"epoch": 0.42886961059291634,
"grad_norm": 84.5,
"learning_rate": 3.068393231845281e-05,
"loss": 34.53681396484375,
"step": 26300
},
{
"epoch": 0.4305002935229274,
"grad_norm": 76.0,
"learning_rate": 3.05587281858343e-05,
"loss": 33.70979248046875,
"step": 26400
},
{
"epoch": 0.43213097645293846,
"grad_norm": 102.0,
"learning_rate": 3.0433377206274887e-05,
"loss": 34.9661328125,
"step": 26500
},
{
"epoch": 0.4337616593829496,
"grad_norm": 77.5,
"learning_rate": 3.0307882691216936e-05,
"loss": 34.26576171875,
"step": 26600
},
{
"epoch": 0.43539234231296065,
"grad_norm": 86.0,
"learning_rate": 3.01822479558946e-05,
"loss": 34.50356201171875,
"step": 26700
},
{
"epoch": 0.43702302524297176,
"grad_norm": 115.0,
"learning_rate": 3.0056476319246308e-05,
"loss": 34.92467041015625,
"step": 26800
},
{
"epoch": 0.4386537081729828,
"grad_norm": 76.0,
"learning_rate": 2.9930571103827047e-05,
"loss": 34.61506103515625,
"step": 26900
},
{
"epoch": 0.44028439110299394,
"grad_norm": 79.5,
"learning_rate": 2.980453563572063e-05,
"loss": 34.48717529296875,
"step": 27000
},
{
"epoch": 0.441915074033005,
"grad_norm": 72.5,
"learning_rate": 2.967837324445175e-05,
"loss": 34.84179443359375,
"step": 27100
},
{
"epoch": 0.4435457569630161,
"grad_norm": 72.5,
"learning_rate": 2.9552087262898148e-05,
"loss": 34.5514794921875,
"step": 27200
},
{
"epoch": 0.4451764398930272,
"grad_norm": 76.5,
"learning_rate": 2.942568102720242e-05,
"loss": 34.153203125,
"step": 27300
},
{
"epoch": 0.4468071228230383,
"grad_norm": 72.5,
"learning_rate": 2.9299157876684e-05,
"loss": 34.22513427734375,
"step": 27400
},
{
"epoch": 0.44843780575304937,
"grad_norm": 73.0,
"learning_rate": 2.917252115375091e-05,
"loss": 34.6579296875,
"step": 27500
},
{
"epoch": 0.4500684886830605,
"grad_norm": 83.5,
"learning_rate": 2.904577420381144e-05,
"loss": 32.9301953125,
"step": 27600
},
{
"epoch": 0.45169917161307155,
"grad_norm": 56.25,
"learning_rate": 2.8918920375185766e-05,
"loss": 33.91248046875,
"step": 27700
},
{
"epoch": 0.45332985454308267,
"grad_norm": 84.5,
"learning_rate": 2.8791963019017564e-05,
"loss": 34.276845703125,
"step": 27800
},
{
"epoch": 0.45496053747309373,
"grad_norm": 81.5,
"learning_rate": 2.8664905489185408e-05,
"loss": 33.42244873046875,
"step": 27900
},
{
"epoch": 0.45659122040310485,
"grad_norm": 86.5,
"learning_rate": 2.853775114221419e-05,
"loss": 34.39563720703125,
"step": 28000
},
{
"epoch": 0.4582219033331159,
"grad_norm": 74.0,
"learning_rate": 2.8410503337186472e-05,
"loss": 34.434716796875,
"step": 28100
},
{
"epoch": 0.459852586263127,
"grad_norm": 88.0,
"learning_rate": 2.8283165435653725e-05,
"loss": 35.30036865234375,
"step": 28200
},
{
"epoch": 0.4614832691931381,
"grad_norm": 82.0,
"learning_rate": 2.8155740801547527e-05,
"loss": 34.15641845703125,
"step": 28300
},
{
"epoch": 0.46311395212314915,
"grad_norm": 66.5,
"learning_rate": 2.8028232801090698e-05,
"loss": 34.61505615234375,
"step": 28400
},
{
"epoch": 0.46474463505316027,
"grad_norm": 80.0,
"learning_rate": 2.7900644802708396e-05,
"loss": 33.96723876953125,
"step": 28500
},
{
"epoch": 0.46637531798317133,
"grad_norm": 72.0,
"learning_rate": 2.77729801769391e-05,
"loss": 34.453193359375,
"step": 28600
},
{
"epoch": 0.46800600091318245,
"grad_norm": 76.5,
"learning_rate": 2.7645242296345592e-05,
"loss": 34.42294921875,
"step": 28700
},
{
"epoch": 0.4696366838431935,
"grad_norm": 122.5,
"learning_rate": 2.7517434535425824e-05,
"loss": 33.4730029296875,
"step": 28800
},
{
"epoch": 0.47126736677320463,
"grad_norm": 78.5,
"learning_rate": 2.7389560270523845e-05,
"loss": 34.14063720703125,
"step": 28900
},
{
"epoch": 0.4728980497032157,
"grad_norm": 78.0,
"learning_rate": 2.726162287974054e-05,
"loss": 34.044091796875,
"step": 29000
},
{
"epoch": 0.4745287326332268,
"grad_norm": 82.0,
"learning_rate": 2.7133625742844404e-05,
"loss": 34.58265869140625,
"step": 29100
},
{
"epoch": 0.4761594155632379,
"grad_norm": 74.0,
"learning_rate": 2.7005572241182292e-05,
"loss": 34.20027587890625,
"step": 29200
},
{
"epoch": 0.477790098493249,
"grad_norm": 79.0,
"learning_rate": 2.6877465757590055e-05,
"loss": 33.4258984375,
"step": 29300
},
{
"epoch": 0.47942078142326006,
"grad_norm": 96.0,
"learning_rate": 2.6749309676303163e-05,
"loss": 34.522021484375,
"step": 29400
},
{
"epoch": 0.4810514643532712,
"grad_norm": 109.5,
"learning_rate": 2.6621107382867384e-05,
"loss": 33.40908935546875,
"step": 29500
},
{
"epoch": 0.48268214728328224,
"grad_norm": 73.5,
"learning_rate": 2.6492862264049227e-05,
"loss": 33.4336376953125,
"step": 29600
},
{
"epoch": 0.48431283021329335,
"grad_norm": 75.5,
"learning_rate": 2.636457770774658e-05,
"loss": 33.89264892578125,
"step": 29700
},
{
"epoch": 0.4859435131433044,
"grad_norm": 98.5,
"learning_rate": 2.6236257102899142e-05,
"loss": 34.32429931640625,
"step": 29800
},
{
"epoch": 0.4875741960733155,
"grad_norm": 85.5,
"learning_rate": 2.6107903839398923e-05,
"loss": 34.54858154296875,
"step": 29900
},
{
"epoch": 0.4892048790033266,
"grad_norm": 85.5,
"learning_rate": 2.5979521308000694e-05,
"loss": 34.4176318359375,
"step": 30000
},
{
"epoch": 0.49083556193333766,
"grad_norm": 75.0,
"learning_rate": 2.5851112900232393e-05,
"loss": 33.9635009765625,
"step": 30100
},
{
"epoch": 0.4924662448633488,
"grad_norm": 85.0,
"learning_rate": 2.5722682008305572e-05,
"loss": 33.53327392578125,
"step": 30200
},
{
"epoch": 0.49409692779335984,
"grad_norm": 80.0,
"learning_rate": 2.559423202502572e-05,
"loss": 33.79457275390625,
"step": 30300
},
{
"epoch": 0.49572761072337096,
"grad_norm": 79.5,
"learning_rate": 2.5465766343702697e-05,
"loss": 34.2709375,
"step": 30400
},
{
"epoch": 0.497358293653382,
"grad_norm": 109.0,
"learning_rate": 2.533728835806105e-05,
"loss": 33.9087548828125,
"step": 30500
},
{
"epoch": 0.49898897658339314,
"grad_norm": 73.0,
"learning_rate": 2.520880146215038e-05,
"loss": 33.56689453125,
"step": 30600
},
{
"epoch": 0.5006196595134043,
"grad_norm": 89.5,
"learning_rate": 2.5080309050255663e-05,
"loss": 33.89188720703125,
"step": 30700
},
{
"epoch": 0.5022503424434153,
"grad_norm": 87.5,
"learning_rate": 2.4951814516807604e-05,
"loss": 35.309658203125,
"step": 30800
},
{
"epoch": 0.5038810253734264,
"grad_norm": 93.5,
"learning_rate": 2.482332125629295e-05,
"loss": 34.22337890625,
"step": 30900
},
{
"epoch": 0.5055117083034375,
"grad_norm": 83.5,
"learning_rate": 2.469483266316483e-05,
"loss": 34.0810546875,
"step": 31000
},
{
"epoch": 0.5071423912334486,
"grad_norm": 80.0,
"learning_rate": 2.456635213175305e-05,
"loss": 33.8587060546875,
"step": 31100
},
{
"epoch": 0.5087730741634596,
"grad_norm": 109.0,
"learning_rate": 2.4437883056174458e-05,
"loss": 35.36123779296875,
"step": 31200
},
{
"epoch": 0.5104037570934707,
"grad_norm": 76.5,
"learning_rate": 2.4309428830243282e-05,
"loss": 33.87925537109375,
"step": 31300
},
{
"epoch": 0.5120344400234819,
"grad_norm": 89.5,
"learning_rate": 2.4180992847381456e-05,
"loss": 34.95505615234375,
"step": 31400
},
{
"epoch": 0.5136651229534929,
"grad_norm": 78.5,
"learning_rate": 2.4052578500528964e-05,
"loss": 34.601640625,
"step": 31500
},
{
"epoch": 0.515295805883504,
"grad_norm": 81.5,
"learning_rate": 2.3924189182054242e-05,
"loss": 33.93820068359375,
"step": 31600
},
{
"epoch": 0.5169264888135151,
"grad_norm": 68.5,
"learning_rate": 2.3795828283664527e-05,
"loss": 34.385,
"step": 31700
},
{
"epoch": 0.5185571717435262,
"grad_norm": 89.5,
"learning_rate": 2.3667499196316288e-05,
"loss": 34.1562109375,
"step": 31800
},
{
"epoch": 0.5201878546735372,
"grad_norm": 91.0,
"learning_rate": 2.35392053101256e-05,
"loss": 33.5622998046875,
"step": 31900
},
{
"epoch": 0.5218185376035483,
"grad_norm": 124.0,
"learning_rate": 2.3410950014278675e-05,
"loss": 33.76296630859375,
"step": 32000
},
{
"epoch": 0.5234492205335595,
"grad_norm": 149.0,
"learning_rate": 2.3282736696942214e-05,
"loss": 33.681259765625,
"step": 32100
},
{
"epoch": 0.5250799034635706,
"grad_norm": 78.0,
"learning_rate": 2.3154568745173982e-05,
"loss": 33.2988330078125,
"step": 32200
},
{
"epoch": 0.5267105863935816,
"grad_norm": 80.0,
"learning_rate": 2.3026449544833294e-05,
"loss": 34.93619873046875,
"step": 32300
},
{
"epoch": 0.5283412693235927,
"grad_norm": 83.5,
"learning_rate": 2.2898382480491584e-05,
"loss": 34.885751953125,
"step": 32400
},
{
"epoch": 0.5299719522536038,
"grad_norm": 96.5,
"learning_rate": 2.2770370935342992e-05,
"loss": 34.28013671875,
"step": 32500
},
{
"epoch": 0.531602635183615,
"grad_norm": 134.0,
"learning_rate": 2.2642418291115007e-05,
"loss": 33.263671875,
"step": 32600
},
{
"epoch": 0.533233318113626,
"grad_norm": 92.5,
"learning_rate": 2.2514527927979068e-05,
"loss": 33.7147802734375,
"step": 32700
},
{
"epoch": 0.5348640010436371,
"grad_norm": 92.0,
"learning_rate": 2.2386703224461354e-05,
"loss": 34.66751220703125,
"step": 32800
},
{
"epoch": 0.5364946839736482,
"grad_norm": 105.0,
"learning_rate": 2.2258947557353456e-05,
"loss": 33.26562255859375,
"step": 32900
},
{
"epoch": 0.5381253669036593,
"grad_norm": 82.0,
"learning_rate": 2.2131264301623226e-05,
"loss": 33.68347412109375,
"step": 33000
},
{
"epoch": 0.5397560498336703,
"grad_norm": 117.0,
"learning_rate": 2.2003656830325602e-05,
"loss": 33.48946044921875,
"step": 33100
},
{
"epoch": 0.5413867327636814,
"grad_norm": 95.0,
"learning_rate": 2.1876128514513473e-05,
"loss": 33.91331298828125,
"step": 33200
},
{
"epoch": 0.5430174156936926,
"grad_norm": 81.0,
"learning_rate": 2.1748682723148675e-05,
"loss": 33.59994140625,
"step": 33300
},
{
"epoch": 0.5446480986237036,
"grad_norm": 97.0,
"learning_rate": 2.1621322823012956e-05,
"loss": 33.696162109375,
"step": 33400
},
{
"epoch": 0.5462787815537147,
"grad_norm": 123.0,
"learning_rate": 2.149405217861904e-05,
"loss": 34.51543701171875,
"step": 33500
},
{
"epoch": 0.5479094644837258,
"grad_norm": 92.5,
"learning_rate": 2.1366874152121753e-05,
"loss": 33.99985107421875,
"step": 33600
},
{
"epoch": 0.5495401474137369,
"grad_norm": 78.5,
"learning_rate": 2.12397921032292e-05,
"loss": 34.8341357421875,
"step": 33700
},
{
"epoch": 0.5511708303437479,
"grad_norm": 76.0,
"learning_rate": 2.1112809389114004e-05,
"loss": 33.99704833984375,
"step": 33800
},
{
"epoch": 0.552801513273759,
"grad_norm": 103.5,
"learning_rate": 2.0985929364324637e-05,
"loss": 33.13259521484375,
"step": 33900
},
{
"epoch": 0.5544321962037702,
"grad_norm": 74.0,
"learning_rate": 2.085915538069679e-05,
"loss": 34.34890625,
"step": 34000
},
{
"epoch": 0.5560628791337813,
"grad_norm": 83.5,
"learning_rate": 2.0732490787264816e-05,
"loss": 34.12480712890625,
"step": 34100
},
{
"epoch": 0.5576935620637923,
"grad_norm": 97.0,
"learning_rate": 2.0605938930173276e-05,
"loss": 34.26929931640625,
"step": 34200
},
{
"epoch": 0.5593242449938034,
"grad_norm": 91.0,
"learning_rate": 2.0479503152588524e-05,
"loss": 33.5249609375,
"step": 34300
},
{
"epoch": 0.5609549279238145,
"grad_norm": 78.5,
"learning_rate": 2.0353186794610412e-05,
"loss": 33.80001220703125,
"step": 34400
},
{
"epoch": 0.5625856108538256,
"grad_norm": 77.0,
"learning_rate": 2.0226993193184023e-05,
"loss": 33.91908935546875,
"step": 34500
},
{
"epoch": 0.5642162937838366,
"grad_norm": 96.5,
"learning_rate": 2.0100925682011564e-05,
"loss": 33.47369384765625,
"step": 34600
},
{
"epoch": 0.5658469767138478,
"grad_norm": 89.0,
"learning_rate": 1.9974987591464254e-05,
"loss": 34.1176806640625,
"step": 34700
},
{
"epoch": 0.5674776596438589,
"grad_norm": 142.0,
"learning_rate": 1.9849182248494344e-05,
"loss": 34.33243896484375,
"step": 34800
},
{
"epoch": 0.5691083425738699,
"grad_norm": 73.0,
"learning_rate": 1.972351297654727e-05,
"loss": 33.624833984375,
"step": 34900
},
{
"epoch": 0.570739025503881,
"grad_norm": 98.0,
"learning_rate": 1.9597983095473825e-05,
"loss": 33.97660888671875,
"step": 35000
},
{
"epoch": 0.5723697084338921,
"grad_norm": 82.5,
"learning_rate": 1.9472595921442444e-05,
"loss": 34.39958740234375,
"step": 35100
},
{
"epoch": 0.5740003913639032,
"grad_norm": 113.5,
"learning_rate": 1.9347354766851637e-05,
"loss": 34.91939697265625,
"step": 35200
},
{
"epoch": 0.5756310742939142,
"grad_norm": 90.5,
"learning_rate": 1.922226294024247e-05,
"loss": 33.7347412109375,
"step": 35300
},
{
"epoch": 0.5772617572239254,
"grad_norm": 99.0,
"learning_rate": 1.909732374621114e-05,
"loss": 33.37375244140625,
"step": 35400
},
{
"epoch": 0.5788924401539365,
"grad_norm": 109.5,
"learning_rate": 1.89725404853217e-05,
"loss": 33.9823681640625,
"step": 35500
},
{
"epoch": 0.5805231230839476,
"grad_norm": 74.5,
"learning_rate": 1.8847916454018875e-05,
"loss": 34.16293701171875,
"step": 35600
},
{
"epoch": 0.5821538060139586,
"grad_norm": 77.5,
"learning_rate": 1.8723454944540936e-05,
"loss": 34.01928955078125,
"step": 35700
},
{
"epoch": 0.5837844889439697,
"grad_norm": 80.0,
"learning_rate": 1.8599159244832777e-05,
"loss": 34.20997314453125,
"step": 35800
},
{
"epoch": 0.5854151718739808,
"grad_norm": 72.5,
"learning_rate": 1.847503263845903e-05,
"loss": 33.38380126953125,
"step": 35900
},
{
"epoch": 0.587045854803992,
"grad_norm": 79.0,
"learning_rate": 1.8351078404517336e-05,
"loss": 34.18912841796875,
"step": 36000
},
{
"epoch": 0.588676537734003,
"grad_norm": 84.0,
"learning_rate": 1.8227299817551686e-05,
"loss": 33.1730712890625,
"step": 36100
},
{
"epoch": 0.5903072206640141,
"grad_norm": 87.0,
"learning_rate": 1.8103700147465967e-05,
"loss": 33.34257080078125,
"step": 36200
},
{
"epoch": 0.5919379035940252,
"grad_norm": 75.5,
"learning_rate": 1.798028265943754e-05,
"loss": 33.53064208984375,
"step": 36300
},
{
"epoch": 0.5935685865240362,
"grad_norm": 113.0,
"learning_rate": 1.7857050613831006e-05,
"loss": 33.52544189453125,
"step": 36400
},
{
"epoch": 0.5951992694540473,
"grad_norm": 93.5,
"learning_rate": 1.7734007266112035e-05,
"loss": 33.58503173828125,
"step": 36500
},
{
"epoch": 0.5968299523840584,
"grad_norm": 78.5,
"learning_rate": 1.7611155866761458e-05,
"loss": 33.90226806640625,
"step": 36600
},
{
"epoch": 0.5984606353140696,
"grad_norm": 75.0,
"learning_rate": 1.7488499661189278e-05,
"loss": 34.0908740234375,
"step": 36700
},
{
"epoch": 0.6000913182440806,
"grad_norm": 90.0,
"learning_rate": 1.7366041889649023e-05,
"loss": 33.82671142578125,
"step": 36800
},
{
"epoch": 0.6017220011740917,
"grad_norm": 78.0,
"learning_rate": 1.72437857871521e-05,
"loss": 33.6208837890625,
"step": 36900
},
{
"epoch": 0.6033526841041028,
"grad_norm": 82.5,
"learning_rate": 1.7121734583382367e-05,
"loss": 33.47661865234375,
"step": 37000
},
{
"epoch": 0.6049833670341139,
"grad_norm": 79.0,
"learning_rate": 1.6999891502610792e-05,
"loss": 33.89511962890625,
"step": 37100
},
{
"epoch": 0.6066140499641249,
"grad_norm": 76.5,
"learning_rate": 1.6878259763610277e-05,
"loss": 34.07520263671875,
"step": 37200
},
{
"epoch": 0.608244732894136,
"grad_norm": 80.5,
"learning_rate": 1.6756842579570636e-05,
"loss": 33.76389892578125,
"step": 37300
},
{
"epoch": 0.6098754158241472,
"grad_norm": 77.5,
"learning_rate": 1.663564315801372e-05,
"loss": 34.45850830078125,
"step": 37400
},
{
"epoch": 0.6115060987541583,
"grad_norm": 92.0,
"learning_rate": 1.6514664700708644e-05,
"loss": 33.12240966796875,
"step": 37500
},
{
"epoch": 0.6131367816841693,
"grad_norm": 86.5,
"learning_rate": 1.639391040358724e-05,
"loss": 34.28488525390625,
"step": 37600
},
{
"epoch": 0.6147674646141804,
"grad_norm": 94.0,
"learning_rate": 1.6273383456659635e-05,
"loss": 33.04328857421875,
"step": 37700
},
{
"epoch": 0.6163981475441915,
"grad_norm": 98.0,
"learning_rate": 1.6153087043929935e-05,
"loss": 33.84422119140625,
"step": 37800
},
{
"epoch": 0.6180288304742027,
"grad_norm": 91.0,
"learning_rate": 1.6033024343312163e-05,
"loss": 33.04386474609375,
"step": 37900
},
{
"epoch": 0.6196595134042137,
"grad_norm": 94.0,
"learning_rate": 1.591319852654628e-05,
"loss": 34.63755615234375,
"step": 38000
},
{
"epoch": 0.6212901963342248,
"grad_norm": 92.5,
"learning_rate": 1.5793612759114397e-05,
"loss": 33.4912255859375,
"step": 38100
},
{
"epoch": 0.6229208792642359,
"grad_norm": 79.0,
"learning_rate": 1.5674270200157146e-05,
"loss": 33.6641796875,
"step": 38200
},
{
"epoch": 0.6245515621942469,
"grad_norm": 71.5,
"learning_rate": 1.5555174002390248e-05,
"loss": 33.8430126953125,
"step": 38300
},
{
"epoch": 0.626182245124258,
"grad_norm": 94.0,
"learning_rate": 1.5436327312021205e-05,
"loss": 33.6026171875,
"step": 38400
},
{
"epoch": 0.6278129280542691,
"grad_norm": 91.0,
"learning_rate": 1.5317733268666173e-05,
"loss": 33.078310546875,
"step": 38500
},
{
"epoch": 0.6294436109842803,
"grad_norm": 86.0,
"learning_rate": 1.5199395005267083e-05,
"loss": 33.33626708984375,
"step": 38600
},
{
"epoch": 0.6310742939142913,
"grad_norm": 88.5,
"learning_rate": 1.50813156480088e-05,
"loss": 33.10360595703125,
"step": 38700
},
{
"epoch": 0.6327049768443024,
"grad_norm": 100.0,
"learning_rate": 1.4963498316236574e-05,
"loss": 34.0476611328125,
"step": 38800
},
{
"epoch": 0.6343356597743135,
"grad_norm": 69.5,
"learning_rate": 1.4845946122373644e-05,
"loss": 33.49295166015625,
"step": 38900
},
{
"epoch": 0.6359663427043246,
"grad_norm": 84.5,
"learning_rate": 1.4728662171838997e-05,
"loss": 33.53701171875,
"step": 39000
},
{
"epoch": 0.6375970256343356,
"grad_norm": 91.5,
"learning_rate": 1.4611649562965332e-05,
"loss": 34.28846435546875,
"step": 39100
},
{
"epoch": 0.6392277085643467,
"grad_norm": 92.0,
"learning_rate": 1.449491138691724e-05,
"loss": 34.3664794921875,
"step": 39200
},
{
"epoch": 0.6408583914943579,
"grad_norm": 82.5,
"learning_rate": 1.4378450727609502e-05,
"loss": 33.4730810546875,
"step": 39300
},
{
"epoch": 0.642489074424369,
"grad_norm": 79.5,
"learning_rate": 1.4262270661625648e-05,
"loss": 33.18058349609375,
"step": 39400
},
{
"epoch": 0.64411975735438,
"grad_norm": 78.5,
"learning_rate": 1.4146374258136653e-05,
"loss": 33.95961181640625,
"step": 39500
},
{
"epoch": 0.6457504402843911,
"grad_norm": 124.5,
"learning_rate": 1.40307645788199e-05,
"loss": 33.35662109375,
"step": 39600
},
{
"epoch": 0.6473811232144022,
"grad_norm": 88.5,
"learning_rate": 1.3915444677778259e-05,
"loss": 33.788955078125,
"step": 39700
},
{
"epoch": 0.6490118061444132,
"grad_norm": 80.5,
"learning_rate": 1.3800417601459433e-05,
"loss": 33.98005615234375,
"step": 39800
},
{
"epoch": 0.6506424890744243,
"grad_norm": 116.0,
"learning_rate": 1.3685686388575469e-05,
"loss": 33.4968798828125,
"step": 39900
},
{
"epoch": 0.6522731720044355,
"grad_norm": 92.0,
"learning_rate": 1.3571254070022471e-05,
"loss": 33.40612548828125,
"step": 40000
},
{
"epoch": 0.6539038549344466,
"grad_norm": 109.5,
"learning_rate": 1.345712366880055e-05,
"loss": 33.54975341796875,
"step": 40100
},
{
"epoch": 0.6555345378644576,
"grad_norm": 102.5,
"learning_rate": 1.334329819993396e-05,
"loss": 33.590986328125,
"step": 40200
},
{
"epoch": 0.6571652207944687,
"grad_norm": 72.0,
"learning_rate": 1.322978067039145e-05,
"loss": 33.3714208984375,
"step": 40300
},
{
"epoch": 0.6587959037244798,
"grad_norm": 71.0,
"learning_rate": 1.31165740790068e-05,
"loss": 33.2227392578125,
"step": 40400
},
{
"epoch": 0.6604265866544909,
"grad_norm": 90.5,
"learning_rate": 1.3003681416399635e-05,
"loss": 33.59010986328125,
"step": 40500
},
{
"epoch": 0.662057269584502,
"grad_norm": 72.0,
"learning_rate": 1.2891105664896442e-05,
"loss": 33.6736181640625,
"step": 40600
},
{
"epoch": 0.6636879525145131,
"grad_norm": 135.0,
"learning_rate": 1.2778849798451714e-05,
"loss": 34.32442626953125,
"step": 40700
},
{
"epoch": 0.6653186354445242,
"grad_norm": 59.0,
"learning_rate": 1.2666916782569433e-05,
"loss": 33.8808447265625,
"step": 40800
},
{
"epoch": 0.6669493183745353,
"grad_norm": 124.5,
"learning_rate": 1.2555309574224715e-05,
"loss": 33.16865234375,
"step": 40900
},
{
"epoch": 0.6685800013045463,
"grad_norm": 79.0,
"learning_rate": 1.2444031121785724e-05,
"loss": 34.5820361328125,
"step": 41000
},
{
"epoch": 0.6702106842345574,
"grad_norm": 82.0,
"learning_rate": 1.2333084364935733e-05,
"loss": 32.6679541015625,
"step": 41100
},
{
"epoch": 0.6718413671645685,
"grad_norm": 95.5,
"learning_rate": 1.2222472234595506e-05,
"loss": 32.66317626953125,
"step": 41200
},
{
"epoch": 0.6734720500945796,
"grad_norm": 100.5,
"learning_rate": 1.2112197652845856e-05,
"loss": 32.90106201171875,
"step": 41300
},
{
"epoch": 0.6751027330245907,
"grad_norm": 72.5,
"learning_rate": 1.2002263532850447e-05,
"loss": 33.786337890625,
"step": 41400
},
{
"epoch": 0.6767334159546018,
"grad_norm": 79.0,
"learning_rate": 1.1892672778778857e-05,
"loss": 33.8490625,
"step": 41500
},
{
"epoch": 0.6783640988846129,
"grad_norm": 81.0,
"learning_rate": 1.1783428285729812e-05,
"loss": 33.3559228515625,
"step": 41600
},
{
"epoch": 0.6799947818146239,
"grad_norm": 101.5,
"learning_rate": 1.1674532939654759e-05,
"loss": 33.4592724609375,
"step": 41700
},
{
"epoch": 0.681625464744635,
"grad_norm": 85.5,
"learning_rate": 1.1565989617281587e-05,
"loss": 33.96642822265625,
"step": 41800
},
{
"epoch": 0.6832561476746462,
"grad_norm": 78.0,
"learning_rate": 1.145780118603868e-05,
"loss": 33.46968505859375,
"step": 41900
},
{
"epoch": 0.6848868306046573,
"grad_norm": 144.0,
"learning_rate": 1.134997050397909e-05,
"loss": 33.69892822265625,
"step": 42000
},
{
"epoch": 0.6865175135346683,
"grad_norm": 103.0,
"learning_rate": 1.1242500419705104e-05,
"loss": 33.31197998046875,
"step": 42100
},
{
"epoch": 0.6881481964646794,
"grad_norm": 86.0,
"learning_rate": 1.113539377229296e-05,
"loss": 32.75554443359375,
"step": 42200
},
{
"epoch": 0.6897788793946905,
"grad_norm": 97.0,
"learning_rate": 1.1028653391217855e-05,
"loss": 34.01078857421875,
"step": 42300
},
{
"epoch": 0.6914095623247016,
"grad_norm": 89.5,
"learning_rate": 1.0922282096279194e-05,
"loss": 33.20049072265625,
"step": 42400
},
{
"epoch": 0.6930402452547126,
"grad_norm": 87.0,
"learning_rate": 1.08162826975261e-05,
"loss": 33.14046875,
"step": 42500
},
{
"epoch": 0.6946709281847238,
"grad_norm": 97.0,
"learning_rate": 1.0710657995183178e-05,
"loss": 33.11598388671875,
"step": 42600
},
{
"epoch": 0.6963016111147349,
"grad_norm": 80.0,
"learning_rate": 1.0605410779576544e-05,
"loss": 33.7512255859375,
"step": 42700
},
{
"epoch": 0.697932294044746,
"grad_norm": 113.5,
"learning_rate": 1.0500543831060116e-05,
"loss": 33.73210693359375,
"step": 42800
},
{
"epoch": 0.699562976974757,
"grad_norm": 90.0,
"learning_rate": 1.0396059919942133e-05,
"loss": 33.78228515625,
"step": 42900
},
{
"epoch": 0.7011936599047681,
"grad_norm": 105.0,
"learning_rate": 1.0291961806412026e-05,
"loss": 33.5402197265625,
"step": 43000
},
{
"epoch": 0.7028243428347792,
"grad_norm": 86.5,
"learning_rate": 1.0188252240467444e-05,
"loss": 33.7085205078125,
"step": 43100
},
{
"epoch": 0.7044550257647902,
"grad_norm": 89.5,
"learning_rate": 1.0084933961841681e-05,
"loss": 33.6503076171875,
"step": 43200
},
{
"epoch": 0.7060857086948014,
"grad_norm": 75.5,
"learning_rate": 9.982009699931187e-06,
"loss": 33.24869384765625,
"step": 43300
},
{
"epoch": 0.7077163916248125,
"grad_norm": 83.0,
"learning_rate": 9.879482173723569e-06,
"loss": 33.42003662109375,
"step": 43400
},
{
"epoch": 0.7093470745548236,
"grad_norm": 124.0,
"learning_rate": 9.777354091725724e-06,
"loss": 33.2013232421875,
"step": 43500
},
{
"epoch": 0.7109777574848346,
"grad_norm": 76.0,
"learning_rate": 9.675628151892272e-06,
"loss": 33.6980859375,
"step": 43600
},
{
"epoch": 0.7126084404148457,
"grad_norm": 96.5,
"learning_rate": 9.57430704155431e-06,
"loss": 33.1621826171875,
"step": 43700
},
{
"epoch": 0.7142391233448568,
"grad_norm": 87.0,
"learning_rate": 9.473393437348373e-06,
"loss": 33.1900048828125,
"step": 43800
},
{
"epoch": 0.715869806274868,
"grad_norm": 123.5,
"learning_rate": 9.372890005145817e-06,
"loss": 33.99087158203125,
"step": 43900
},
{
"epoch": 0.717500489204879,
"grad_norm": 89.0,
"learning_rate": 9.272799399982301e-06,
"loss": 34.00809326171875,
"step": 44000
},
{
"epoch": 0.7191311721348901,
"grad_norm": 112.0,
"learning_rate": 9.17312426598767e-06,
"loss": 34.3621142578125,
"step": 44100
},
{
"epoch": 0.7207618550649012,
"grad_norm": 86.0,
"learning_rate": 9.073867236316135e-06,
"loss": 34.213896484375,
"step": 44200
},
{
"epoch": 0.7223925379949123,
"grad_norm": 108.5,
"learning_rate": 8.975030933076695e-06,
"loss": 33.28839111328125,
"step": 44300
},
{
"epoch": 0.7240232209249233,
"grad_norm": 106.0,
"learning_rate": 8.876617967263855e-06,
"loss": 34.2870166015625,
"step": 44400
},
{
"epoch": 0.7256539038549344,
"grad_norm": 89.0,
"learning_rate": 8.778630938688676e-06,
"loss": 33.61921630859375,
"step": 44500
},
{
"epoch": 0.7272845867849456,
"grad_norm": 85.0,
"learning_rate": 8.681072435910062e-06,
"loss": 33.1503076171875,
"step": 44600
},
{
"epoch": 0.7289152697149566,
"grad_norm": 86.5,
"learning_rate": 8.58394503616641e-06,
"loss": 33.38920654296875,
"step": 44700
},
{
"epoch": 0.7305459526449677,
"grad_norm": 77.5,
"learning_rate": 8.487251305307507e-06,
"loss": 33.65532470703125,
"step": 44800
},
{
"epoch": 0.7321766355749788,
"grad_norm": 111.5,
"learning_rate": 8.390993797726757e-06,
"loss": 34.0880224609375,
"step": 44900
},
{
"epoch": 0.7338073185049899,
"grad_norm": 89.0,
"learning_rate": 8.295175056293676e-06,
"loss": 33.722509765625,
"step": 45000
},
{
"epoch": 0.7354380014350009,
"grad_norm": 88.0,
"learning_rate": 8.199797612286744e-06,
"loss": 33.61972412109375,
"step": 45100
},
{
"epoch": 0.737068684365012,
"grad_norm": 94.0,
"learning_rate": 8.104863985326553e-06,
"loss": 33.24302001953125,
"step": 45200
},
{
"epoch": 0.7386993672950232,
"grad_norm": 85.5,
"learning_rate": 8.010376683309193e-06,
"loss": 34.3585888671875,
"step": 45300
},
{
"epoch": 0.7403300502250343,
"grad_norm": 79.5,
"learning_rate": 7.916338202340018e-06,
"loss": 33.6855224609375,
"step": 45400
},
{
"epoch": 0.7419607331550453,
"grad_norm": 92.0,
"learning_rate": 7.82275102666773e-06,
"loss": 33.926123046875,
"step": 45500
},
{
"epoch": 0.7435914160850564,
"grad_norm": 91.0,
"learning_rate": 7.729617628618735e-06,
"loss": 33.21872314453125,
"step": 45600
},
{
"epoch": 0.7452220990150675,
"grad_norm": 82.0,
"learning_rate": 7.636940468531822e-06,
"loss": 33.9178173828125,
"step": 45700
},
{
"epoch": 0.7468527819450786,
"grad_norm": 115.0,
"learning_rate": 7.544721994693185e-06,
"loss": 33.60979736328125,
"step": 45800
},
{
"epoch": 0.7484834648750897,
"grad_norm": 143.0,
"learning_rate": 7.452964643271732e-06,
"loss": 33.27378173828125,
"step": 45900
},
{
"epoch": 0.7501141478051008,
"grad_norm": 96.5,
"learning_rate": 7.361670838254728e-06,
"loss": 33.7872998046875,
"step": 46000
},
{
"epoch": 0.7517448307351119,
"grad_norm": 77.0,
"learning_rate": 7.2708429913837835e-06,
"loss": 33.98447509765625,
"step": 46100
},
{
"epoch": 0.753375513665123,
"grad_norm": 88.0,
"learning_rate": 7.18048350209109e-06,
"loss": 33.3944970703125,
"step": 46200
},
{
"epoch": 0.755006196595134,
"grad_norm": 75.0,
"learning_rate": 7.090594757436095e-06,
"loss": 33.1852294921875,
"step": 46300
},
{
"epoch": 0.7566368795251451,
"grad_norm": 110.5,
"learning_rate": 7.001179132042398e-06,
"loss": 33.26583740234375,
"step": 46400
},
{
"epoch": 0.7582675624551563,
"grad_norm": 72.5,
"learning_rate": 6.912238988035066e-06,
"loss": 33.809775390625,
"step": 46500
},
{
"epoch": 0.7598982453851673,
"grad_norm": 82.5,
"learning_rate": 6.823776674978155e-06,
"loss": 34.18914306640625,
"step": 46600
},
{
"epoch": 0.7615289283151784,
"grad_norm": 85.5,
"learning_rate": 6.735794529812714e-06,
"loss": 33.7421484375,
"step": 46700
},
{
"epoch": 0.7631596112451895,
"grad_norm": 93.0,
"learning_rate": 6.648294876795022e-06,
"loss": 33.18773681640625,
"step": 46800
},
{
"epoch": 0.7647902941752006,
"grad_norm": 80.0,
"learning_rate": 6.5612800274351876e-06,
"loss": 33.82040771484375,
"step": 46900
},
{
"epoch": 0.7664209771052116,
"grad_norm": 87.5,
"learning_rate": 6.474752280436083e-06,
"loss": 33.65014892578125,
"step": 47000
},
{
"epoch": 0.7680516600352227,
"grad_norm": 76.0,
"learning_rate": 2.4750000000000002e-05,
"loss": 33.636025390625,
"step": 47100
},
{
"epoch": 0.7696823429652339,
"grad_norm": 77.0,
"learning_rate": 4.975e-05,
"loss": 33.99931640625,
"step": 47200
},
{
"epoch": 0.771313025895245,
"grad_norm": 90.5,
"learning_rate": 4.9999676354266065e-05,
"loss": 35.69708984375,
"step": 47300
},
{
"epoch": 0.772943708825256,
"grad_norm": 80.5,
"learning_rate": 4.999869231602613e-05,
"loss": 36.07453369140625,
"step": 47400
},
{
"epoch": 0.7745743917552671,
"grad_norm": 88.0,
"learning_rate": 4.999704787825411e-05,
"loss": 37.15786376953125,
"step": 47500
},
{
"epoch": 0.7762050746852782,
"grad_norm": 92.5,
"learning_rate": 4.9994743084391704e-05,
"loss": 36.407890625,
"step": 47600
},
{
"epoch": 0.7778357576152893,
"grad_norm": 74.5,
"learning_rate": 4.99917779953255e-05,
"loss": 35.9034375,
"step": 47700
},
{
"epoch": 0.7794664405453003,
"grad_norm": 68.5,
"learning_rate": 4.998815268938533e-05,
"loss": 36.46909423828125,
"step": 47800
},
{
"epoch": 0.7810971234753115,
"grad_norm": 77.5,
"learning_rate": 4.998386726234221e-05,
"loss": 37.04531982421875,
"step": 47900
},
{
"epoch": 0.7827278064053226,
"grad_norm": 76.0,
"learning_rate": 4.997892182740583e-05,
"loss": 36.817294921875,
"step": 48000
},
{
"epoch": 0.7843584893353336,
"grad_norm": 72.5,
"learning_rate": 4.997331651522155e-05,
"loss": 36.5568359375,
"step": 48100
},
{
"epoch": 0.7859891722653447,
"grad_norm": 84.0,
"learning_rate": 4.996705147386691e-05,
"loss": 36.517900390625,
"step": 48200
},
{
"epoch": 0.7876198551953558,
"grad_norm": 69.5,
"learning_rate": 4.996012686884781e-05,
"loss": 36.15031494140625,
"step": 48300
},
{
"epoch": 0.7892505381253669,
"grad_norm": 92.5,
"learning_rate": 4.9952542883094035e-05,
"loss": 35.585830078125,
"step": 48400
},
{
"epoch": 0.790881221055378,
"grad_norm": 74.0,
"learning_rate": 4.9944299716954504e-05,
"loss": 36.430771484375,
"step": 48500
},
{
"epoch": 0.7925119039853891,
"grad_norm": 74.0,
"learning_rate": 4.9935397588191925e-05,
"loss": 36.07919677734375,
"step": 48600
},
{
"epoch": 0.7941425869154002,
"grad_norm": 73.5,
"learning_rate": 4.992583673197707e-05,
"loss": 35.98107421875,
"step": 48700
},
{
"epoch": 0.7957732698454113,
"grad_norm": 66.5,
"learning_rate": 4.9915617400882534e-05,
"loss": 36.59056884765625,
"step": 48800
},
{
"epoch": 0.7974039527754223,
"grad_norm": 72.0,
"learning_rate": 4.990473986487611e-05,
"loss": 36.108662109375,
"step": 48900
},
{
"epoch": 0.7990346357054334,
"grad_norm": 58.75,
"learning_rate": 4.989320441131363e-05,
"loss": 35.82669921875,
"step": 49000
},
{
"epoch": 0.8006653186354445,
"grad_norm": 62.25,
"learning_rate": 4.9881011344931334e-05,
"loss": 35.23743408203125,
"step": 49100
},
{
"epoch": 0.8022960015654557,
"grad_norm": 71.0,
"learning_rate": 4.98681609878379e-05,
"loss": 36.0745703125,
"step": 49200
},
{
"epoch": 0.8039266844954667,
"grad_norm": 55.75,
"learning_rate": 4.9854653679505856e-05,
"loss": 35.5943212890625,
"step": 49300
},
{
"epoch": 0.8055573674254778,
"grad_norm": 82.0,
"learning_rate": 4.98404897767627e-05,
"loss": 35.40338134765625,
"step": 49400
},
{
"epoch": 0.8071880503554889,
"grad_norm": 73.5,
"learning_rate": 4.982566965378138e-05,
"loss": 36.7010595703125,
"step": 49500
},
{
"epoch": 0.8088187332854999,
"grad_norm": 77.5,
"learning_rate": 4.981019370207046e-05,
"loss": 36.1892236328125,
"step": 49600
},
{
"epoch": 0.810449416215511,
"grad_norm": 61.75,
"learning_rate": 4.9794062330463786e-05,
"loss": 36.35080322265625,
"step": 49700
},
{
"epoch": 0.8120800991455221,
"grad_norm": 82.5,
"learning_rate": 4.977727596510966e-05,
"loss": 36.4447509765625,
"step": 49800
},
{
"epoch": 0.8137107820755333,
"grad_norm": 71.5,
"learning_rate": 4.975983504945958e-05,
"loss": 35.958818359375,
"step": 49900
},
{
"epoch": 0.8153414650055443,
"grad_norm": 66.0,
"learning_rate": 4.974174004425656e-05,
"loss": 35.521708984375,
"step": 50000
},
{
"epoch": 0.8169721479355554,
"grad_norm": 76.5,
"learning_rate": 4.9722991427522906e-05,
"loss": 36.17392333984375,
"step": 50100
},
{
"epoch": 0.8186028308655665,
"grad_norm": 59.0,
"learning_rate": 4.970358969454765e-05,
"loss": 35.87954345703125,
"step": 50200
},
{
"epoch": 0.8202335137955776,
"grad_norm": 52.75,
"learning_rate": 4.9683535357873414e-05,
"loss": 35.6118359375,
"step": 50300
},
{
"epoch": 0.8218641967255886,
"grad_norm": 82.0,
"learning_rate": 4.966282894728288e-05,
"loss": 35.3713623046875,
"step": 50400
},
{
"epoch": 0.8234948796555998,
"grad_norm": 62.75,
"learning_rate": 4.964147100978483e-05,
"loss": 35.97928466796875,
"step": 50500
},
{
"epoch": 0.8251255625856109,
"grad_norm": 66.5,
"learning_rate": 4.961946210959965e-05,
"loss": 36.159267578125,
"step": 50600
},
{
"epoch": 0.826756245515622,
"grad_norm": 61.25,
"learning_rate": 4.959680282814445e-05,
"loss": 34.92261474609375,
"step": 50700
},
{
"epoch": 0.828386928445633,
"grad_norm": 62.5,
"learning_rate": 4.9573493764017684e-05,
"loss": 35.4424267578125,
"step": 50800
},
{
"epoch": 0.8300176113756441,
"grad_norm": 71.5,
"learning_rate": 4.954953553298337e-05,
"loss": 35.15792236328125,
"step": 50900
},
{
"epoch": 0.8316482943056552,
"grad_norm": 60.25,
"learning_rate": 4.9524928767954806e-05,
"loss": 35.02305908203125,
"step": 51000
},
{
"epoch": 0.8332789772356664,
"grad_norm": 75.5,
"learning_rate": 4.9499674118977824e-05,
"loss": 35.4053857421875,
"step": 51100
},
{
"epoch": 0.8349096601656774,
"grad_norm": 56.25,
"learning_rate": 4.947377225321366e-05,
"loss": 35.6231787109375,
"step": 51200
},
{
"epoch": 0.8365403430956885,
"grad_norm": 54.25,
"learning_rate": 4.944722385492132e-05,
"loss": 35.4303173828125,
"step": 51300
},
{
"epoch": 0.8381710260256996,
"grad_norm": 68.5,
"learning_rate": 4.9420029625439456e-05,
"loss": 34.95726806640625,
"step": 51400
},
{
"epoch": 0.8398017089557106,
"grad_norm": 60.0,
"learning_rate": 4.939219028316793e-05,
"loss": 36.3287060546875,
"step": 51500
},
{
"epoch": 0.8414323918857217,
"grad_norm": 60.5,
"learning_rate": 4.936370656354873e-05,
"loss": 35.6472314453125,
"step": 51600
},
{
"epoch": 0.8430630748157328,
"grad_norm": 75.5,
"learning_rate": 4.9334579219046646e-05,
"loss": 35.49300048828125,
"step": 51700
},
{
"epoch": 0.844693757745744,
"grad_norm": 64.5,
"learning_rate": 4.930480901912929e-05,
"loss": 34.67722412109375,
"step": 51800
},
{
"epoch": 0.846324440675755,
"grad_norm": 79.5,
"learning_rate": 4.927439675024686e-05,
"loss": 35.6264892578125,
"step": 51900
},
{
"epoch": 0.8479551236057661,
"grad_norm": 76.5,
"learning_rate": 4.92433432158113e-05,
"loss": 35.258955078125,
"step": 52000
},
{
"epoch": 0.8495858065357772,
"grad_norm": 69.0,
"learning_rate": 4.92116492361751e-05,
"loss": 36.43853271484375,
"step": 52100
},
{
"epoch": 0.8512164894657883,
"grad_norm": 77.5,
"learning_rate": 4.917931564860964e-05,
"loss": 34.96982421875,
"step": 52200
},
{
"epoch": 0.8528471723957993,
"grad_norm": 72.5,
"learning_rate": 4.9146343307283026e-05,
"loss": 34.39767578125,
"step": 52300
},
{
"epoch": 0.8544778553258104,
"grad_norm": 70.0,
"learning_rate": 4.911273308323758e-05,
"loss": 34.4587353515625,
"step": 52400
},
{
"epoch": 0.8561085382558216,
"grad_norm": 70.5,
"learning_rate": 4.9078485864366806e-05,
"loss": 35.28226806640625,
"step": 52500
},
{
"epoch": 0.8577392211858327,
"grad_norm": 66.0,
"learning_rate": 4.9043602555391905e-05,
"loss": 35.18193359375,
"step": 52600
},
{
"epoch": 0.8593699041158437,
"grad_norm": 65.0,
"learning_rate": 4.900808407783792e-05,
"loss": 35.25165283203125,
"step": 52700
},
{
"epoch": 0.8610005870458548,
"grad_norm": 70.5,
"learning_rate": 4.897193137000938e-05,
"loss": 35.25130126953125,
"step": 52800
},
{
"epoch": 0.8626312699758659,
"grad_norm": 82.0,
"learning_rate": 4.893514538696547e-05,
"loss": 35.08494384765625,
"step": 52900
},
{
"epoch": 0.8642619529058769,
"grad_norm": 74.0,
"learning_rate": 4.8897727100494894e-05,
"loss": 35.11744873046875,
"step": 53000
},
{
"epoch": 0.865892635835888,
"grad_norm": 73.5,
"learning_rate": 4.885967749909009e-05,
"loss": 34.72285888671875,
"step": 53100
},
{
"epoch": 0.8675233187658992,
"grad_norm": 60.75,
"learning_rate": 4.88209975879212e-05,
"loss": 35.22382568359375,
"step": 53200
},
{
"epoch": 0.8691540016959103,
"grad_norm": 67.5,
"learning_rate": 4.878168838880948e-05,
"loss": 34.6228662109375,
"step": 53300
},
{
"epoch": 0.8707846846259213,
"grad_norm": 53.0,
"learning_rate": 4.874175094020033e-05,
"loss": 33.85436767578125,
"step": 53400
},
{
"epoch": 0.8724153675559324,
"grad_norm": 88.0,
"learning_rate": 4.8701186297135806e-05,
"loss": 34.3983154296875,
"step": 53500
},
{
"epoch": 0.8740460504859435,
"grad_norm": 60.5,
"learning_rate": 4.8659995531226835e-05,
"loss": 34.58441162109375,
"step": 53600
},
{
"epoch": 0.8756767334159546,
"grad_norm": 67.5,
"learning_rate": 4.861817973062483e-05,
"loss": 34.25955810546875,
"step": 53700
},
{
"epoch": 0.8773074163459657,
"grad_norm": 65.5,
"learning_rate": 4.857573999999297e-05,
"loss": 35.13526123046875,
"step": 53800
},
{
"epoch": 0.8789380992759768,
"grad_norm": 83.0,
"learning_rate": 4.853267746047704e-05,
"loss": 34.0967041015625,
"step": 53900
},
{
"epoch": 0.8805687822059879,
"grad_norm": 84.5,
"learning_rate": 4.8488993249675786e-05,
"loss": 34.4576806640625,
"step": 54000
},
{
"epoch": 0.882199465135999,
"grad_norm": 68.5,
"learning_rate": 4.844468852161086e-05,
"loss": 34.2941162109375,
"step": 54100
},
{
"epoch": 0.88383014806601,
"grad_norm": 61.25,
"learning_rate": 4.839976444669634e-05,
"loss": 35.0467724609375,
"step": 54200
},
{
"epoch": 0.8854608309960211,
"grad_norm": 66.0,
"learning_rate": 4.835422221170785e-05,
"loss": 35.232822265625,
"step": 54300
},
{
"epoch": 0.8870915139260322,
"grad_norm": 68.5,
"learning_rate": 4.8308063019751126e-05,
"loss": 34.10639404296875,
"step": 54400
},
{
"epoch": 0.8887221968560433,
"grad_norm": 64.5,
"learning_rate": 4.826128809023032e-05,
"loss": 34.97212890625,
"step": 54500
},
{
"epoch": 0.8903528797860544,
"grad_norm": 76.0,
"learning_rate": 4.821389865881571e-05,
"loss": 34.4078857421875,
"step": 54600
},
{
"epoch": 0.8919835627160655,
"grad_norm": 76.5,
"learning_rate": 4.816589597741114e-05,
"loss": 33.82409423828125,
"step": 54700
},
{
"epoch": 0.8936142456460766,
"grad_norm": 66.0,
"learning_rate": 4.811728131412085e-05,
"loss": 34.2156005859375,
"step": 54800
},
{
"epoch": 0.8952449285760876,
"grad_norm": 61.25,
"learning_rate": 4.806805595321606e-05,
"loss": 35.0901611328125,
"step": 54900
},
{
"epoch": 0.8968756115060987,
"grad_norm": 82.0,
"learning_rate": 4.8018221195101007e-05,
"loss": 34.81978515625,
"step": 55000
},
{
"epoch": 0.8985062944361099,
"grad_norm": 69.5,
"learning_rate": 4.796777835627858e-05,
"loss": 34.64293212890625,
"step": 55100
},
{
"epoch": 0.900136977366121,
"grad_norm": 78.0,
"learning_rate": 4.7916728769315585e-05,
"loss": 34.4116552734375,
"step": 55200
},
{
"epoch": 0.901767660296132,
"grad_norm": 70.5,
"learning_rate": 4.786507378280749e-05,
"loss": 33.916279296875,
"step": 55300
},
{
"epoch": 0.9033983432261431,
"grad_norm": 63.5,
"learning_rate": 4.781281476134283e-05,
"loss": 34.0052880859375,
"step": 55400
},
{
"epoch": 0.9050290261561542,
"grad_norm": 81.0,
"learning_rate": 4.7759953085467144e-05,
"loss": 34.62841552734375,
"step": 55500
},
{
"epoch": 0.9066597090861653,
"grad_norm": 88.5,
"learning_rate": 4.7706490151646525e-05,
"loss": 33.8224951171875,
"step": 55600
},
{
"epoch": 0.9082903920161763,
"grad_norm": 67.0,
"learning_rate": 4.7652427372230714e-05,
"loss": 34.18286865234375,
"step": 55700
},
{
"epoch": 0.9099210749461875,
"grad_norm": 67.0,
"learning_rate": 4.759776617541577e-05,
"loss": 34.788623046875,
"step": 55800
},
{
"epoch": 0.9115517578761986,
"grad_norm": 82.0,
"learning_rate": 4.754250800520639e-05,
"loss": 34.23928955078125,
"step": 55900
},
{
"epoch": 0.9131824408062097,
"grad_norm": 80.0,
"learning_rate": 4.748665432137771e-05,
"loss": 34.2428173828125,
"step": 56000
},
{
"epoch": 0.9148131237362207,
"grad_norm": 80.5,
"learning_rate": 4.743020659943679e-05,
"loss": 34.38558349609375,
"step": 56100
},
{
"epoch": 0.9164438066662318,
"grad_norm": 71.5,
"learning_rate": 4.7373166330583586e-05,
"loss": 34.04652587890625,
"step": 56200
},
{
"epoch": 0.9180744895962429,
"grad_norm": 78.0,
"learning_rate": 4.7315535021671596e-05,
"loss": 33.941416015625,
"step": 56300
},
{
"epoch": 0.919705172526254,
"grad_norm": 63.0,
"learning_rate": 4.725731419516804e-05,
"loss": 34.36150634765625,
"step": 56400
},
{
"epoch": 0.9213358554562651,
"grad_norm": 77.0,
"learning_rate": 4.719850538911361e-05,
"loss": 34.08754150390625,
"step": 56500
},
{
"epoch": 0.9229665383862762,
"grad_norm": 88.5,
"learning_rate": 4.7139110157081926e-05,
"loss": 35.2251318359375,
"step": 56600
},
{
"epoch": 0.9245972213162873,
"grad_norm": 63.75,
"learning_rate": 4.707913006813836e-05,
"loss": 33.99910400390625,
"step": 56700
},
{
"epoch": 0.9262279042462983,
"grad_norm": 69.0,
"learning_rate": 4.7018566706798726e-05,
"loss": 33.97303955078125,
"step": 56800
},
{
"epoch": 0.9278585871763094,
"grad_norm": 68.0,
"learning_rate": 4.695742167298733e-05,
"loss": 33.24641845703125,
"step": 56900
},
{
"epoch": 0.9294892701063205,
"grad_norm": 67.0,
"learning_rate": 4.689569658199472e-05,
"loss": 33.81060546875,
"step": 57000
}
],
"logging_steps": 100,
"max_steps": 61323,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 1000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 8.635314972999118e+17,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}