agent_random_adapter / trainer_state.json
wonwonn's picture
Upload LoRA adapter (Qwen2.5-VL-7B-sft-random)
d7b41a0 verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"eval_steps": 500,
"global_step": 1662,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0012040939193257074,
"grad_norm": 1.0888201950123075,
"learning_rate": 0.0,
"loss": 1.2516779899597168,
"step": 1
},
{
"epoch": 0.002408187838651415,
"grad_norm": 1.3504171031717647,
"learning_rate": 1.1904761904761904e-06,
"loss": 1.292458415031433,
"step": 2
},
{
"epoch": 0.003612281757977122,
"grad_norm": 1.3295966510635613,
"learning_rate": 2.3809523809523808e-06,
"loss": 1.2252806425094604,
"step": 3
},
{
"epoch": 0.00481637567730283,
"grad_norm": 1.2049241506681116,
"learning_rate": 3.5714285714285714e-06,
"loss": 1.1460682153701782,
"step": 4
},
{
"epoch": 0.006020469596628537,
"grad_norm": 1.3299344704725482,
"learning_rate": 4.7619047619047615e-06,
"loss": 1.3050752878189087,
"step": 5
},
{
"epoch": 0.007224563515954244,
"grad_norm": 1.1937833887885951,
"learning_rate": 5.9523809523809525e-06,
"loss": 1.1561931371688843,
"step": 6
},
{
"epoch": 0.008428657435279952,
"grad_norm": 1.1972161646097195,
"learning_rate": 7.142857142857143e-06,
"loss": 1.348860502243042,
"step": 7
},
{
"epoch": 0.00963275135460566,
"grad_norm": 1.1599119044275834,
"learning_rate": 8.333333333333334e-06,
"loss": 1.1653285026550293,
"step": 8
},
{
"epoch": 0.010836845273931367,
"grad_norm": 1.168880883596748,
"learning_rate": 9.523809523809523e-06,
"loss": 1.2427582740783691,
"step": 9
},
{
"epoch": 0.012040939193257074,
"grad_norm": 1.1907331737748812,
"learning_rate": 1.0714285714285714e-05,
"loss": 1.1958000659942627,
"step": 10
},
{
"epoch": 0.013245033112582781,
"grad_norm": 1.0764330357862444,
"learning_rate": 1.1904761904761905e-05,
"loss": 1.0088632106781006,
"step": 11
},
{
"epoch": 0.014449127031908489,
"grad_norm": 0.970590293533405,
"learning_rate": 1.3095238095238096e-05,
"loss": 1.057112455368042,
"step": 12
},
{
"epoch": 0.015653220951234198,
"grad_norm": 0.9117577162508362,
"learning_rate": 1.4285714285714285e-05,
"loss": 1.093891978263855,
"step": 13
},
{
"epoch": 0.016857314870559904,
"grad_norm": 0.7128858566790944,
"learning_rate": 1.5476190476190476e-05,
"loss": 0.9809948205947876,
"step": 14
},
{
"epoch": 0.018061408789885613,
"grad_norm": 0.7466873900766856,
"learning_rate": 1.6666666666666667e-05,
"loss": 1.1132748126983643,
"step": 15
},
{
"epoch": 0.01926550270921132,
"grad_norm": 0.758857309279139,
"learning_rate": 1.785714285714286e-05,
"loss": 0.9380471706390381,
"step": 16
},
{
"epoch": 0.020469596628537028,
"grad_norm": 0.6736348254664769,
"learning_rate": 1.9047619047619046e-05,
"loss": 1.1095030307769775,
"step": 17
},
{
"epoch": 0.021673690547862733,
"grad_norm": 0.675335552277212,
"learning_rate": 2.023809523809524e-05,
"loss": 1.0808950662612915,
"step": 18
},
{
"epoch": 0.022877784467188442,
"grad_norm": 0.7264824733143365,
"learning_rate": 2.1428571428571428e-05,
"loss": 0.9620448350906372,
"step": 19
},
{
"epoch": 0.024081878386514148,
"grad_norm": 0.7051338547341263,
"learning_rate": 2.261904761904762e-05,
"loss": 0.9811359643936157,
"step": 20
},
{
"epoch": 0.025285972305839857,
"grad_norm": 0.7843003159004286,
"learning_rate": 2.380952380952381e-05,
"loss": 1.0925147533416748,
"step": 21
},
{
"epoch": 0.026490066225165563,
"grad_norm": 0.7516975231625368,
"learning_rate": 2.5e-05,
"loss": 0.9252520799636841,
"step": 22
},
{
"epoch": 0.027694160144491272,
"grad_norm": 0.6962403835861913,
"learning_rate": 2.6190476190476192e-05,
"loss": 1.0134961605072021,
"step": 23
},
{
"epoch": 0.028898254063816978,
"grad_norm": 0.7671373768051949,
"learning_rate": 2.7380952380952383e-05,
"loss": 1.0397887229919434,
"step": 24
},
{
"epoch": 0.030102347983142687,
"grad_norm": 0.7085458224295194,
"learning_rate": 2.857142857142857e-05,
"loss": 1.0200514793395996,
"step": 25
},
{
"epoch": 0.031306441902468396,
"grad_norm": 0.6208311360775705,
"learning_rate": 2.9761904761904762e-05,
"loss": 0.8722552061080933,
"step": 26
},
{
"epoch": 0.0325105358217941,
"grad_norm": 0.7245939746288178,
"learning_rate": 3.095238095238095e-05,
"loss": 1.0895636081695557,
"step": 27
},
{
"epoch": 0.03371462974111981,
"grad_norm": 0.7155183357450243,
"learning_rate": 3.2142857142857144e-05,
"loss": 0.9978558421134949,
"step": 28
},
{
"epoch": 0.034918723660445516,
"grad_norm": 0.7124130674118848,
"learning_rate": 3.3333333333333335e-05,
"loss": 1.0731256008148193,
"step": 29
},
{
"epoch": 0.036122817579771226,
"grad_norm": 0.6936933307418388,
"learning_rate": 3.4523809523809526e-05,
"loss": 1.0017905235290527,
"step": 30
},
{
"epoch": 0.03732691149909693,
"grad_norm": 0.6914756992106068,
"learning_rate": 3.571428571428572e-05,
"loss": 0.9039508700370789,
"step": 31
},
{
"epoch": 0.03853100541842264,
"grad_norm": 0.7530165617199812,
"learning_rate": 3.690476190476191e-05,
"loss": 1.018087387084961,
"step": 32
},
{
"epoch": 0.039735099337748346,
"grad_norm": 0.7253115434648741,
"learning_rate": 3.809523809523809e-05,
"loss": 0.9909164309501648,
"step": 33
},
{
"epoch": 0.040939193257074055,
"grad_norm": 0.6415639959381287,
"learning_rate": 3.928571428571429e-05,
"loss": 0.9909334182739258,
"step": 34
},
{
"epoch": 0.04214328717639976,
"grad_norm": 0.6628002571565351,
"learning_rate": 4.047619047619048e-05,
"loss": 0.8408276438713074,
"step": 35
},
{
"epoch": 0.04334738109572547,
"grad_norm": 0.6544680745811386,
"learning_rate": 4.166666666666667e-05,
"loss": 0.9553419947624207,
"step": 36
},
{
"epoch": 0.044551475015051176,
"grad_norm": 0.7501400971456818,
"learning_rate": 4.2857142857142856e-05,
"loss": 0.991558313369751,
"step": 37
},
{
"epoch": 0.045755568934376885,
"grad_norm": 0.6299802150920484,
"learning_rate": 4.404761904761905e-05,
"loss": 0.880527138710022,
"step": 38
},
{
"epoch": 0.04695966285370259,
"grad_norm": 0.7119587435770088,
"learning_rate": 4.523809523809524e-05,
"loss": 1.084947109222412,
"step": 39
},
{
"epoch": 0.048163756773028296,
"grad_norm": 0.6210760615390281,
"learning_rate": 4.642857142857143e-05,
"loss": 0.8718366622924805,
"step": 40
},
{
"epoch": 0.049367850692354005,
"grad_norm": 0.6905357174119584,
"learning_rate": 4.761904761904762e-05,
"loss": 1.030173897743225,
"step": 41
},
{
"epoch": 0.050571944611679714,
"grad_norm": 0.6604045749465294,
"learning_rate": 4.880952380952381e-05,
"loss": 0.8225071430206299,
"step": 42
},
{
"epoch": 0.05177603853100542,
"grad_norm": 0.7288884168617961,
"learning_rate": 5e-05,
"loss": 0.8640418648719788,
"step": 43
},
{
"epoch": 0.052980132450331126,
"grad_norm": 0.7530293725220191,
"learning_rate": 5.119047619047619e-05,
"loss": 1.053181529045105,
"step": 44
},
{
"epoch": 0.054184226369656835,
"grad_norm": 0.633690102398682,
"learning_rate": 5.2380952380952384e-05,
"loss": 0.9794737100601196,
"step": 45
},
{
"epoch": 0.055388320288982544,
"grad_norm": 0.6366432445096702,
"learning_rate": 5.3571428571428575e-05,
"loss": 0.928667426109314,
"step": 46
},
{
"epoch": 0.056592414208308246,
"grad_norm": 0.6635606705213237,
"learning_rate": 5.4761904761904766e-05,
"loss": 0.8890948295593262,
"step": 47
},
{
"epoch": 0.057796508127633955,
"grad_norm": 0.717685283959276,
"learning_rate": 5.595238095238096e-05,
"loss": 0.9435674548149109,
"step": 48
},
{
"epoch": 0.059000602046959665,
"grad_norm": 0.672785636669423,
"learning_rate": 5.714285714285714e-05,
"loss": 0.8677853345870972,
"step": 49
},
{
"epoch": 0.060204695966285374,
"grad_norm": 0.6736728035828868,
"learning_rate": 5.833333333333334e-05,
"loss": 0.904883861541748,
"step": 50
},
{
"epoch": 0.061408789885611076,
"grad_norm": 0.6832027265384513,
"learning_rate": 5.9523809523809524e-05,
"loss": 0.8402994871139526,
"step": 51
},
{
"epoch": 0.06261288380493679,
"grad_norm": 0.6333608523885448,
"learning_rate": 6.0714285714285715e-05,
"loss": 0.8128600716590881,
"step": 52
},
{
"epoch": 0.0638169777242625,
"grad_norm": 0.6511799389683816,
"learning_rate": 6.19047619047619e-05,
"loss": 0.9272807836532593,
"step": 53
},
{
"epoch": 0.0650210716435882,
"grad_norm": 0.7660958513490825,
"learning_rate": 6.30952380952381e-05,
"loss": 0.9557014107704163,
"step": 54
},
{
"epoch": 0.06622516556291391,
"grad_norm": 0.6336813527780781,
"learning_rate": 6.428571428571429e-05,
"loss": 0.9378836154937744,
"step": 55
},
{
"epoch": 0.06742925948223961,
"grad_norm": 0.7508486475685524,
"learning_rate": 6.547619047619048e-05,
"loss": 0.8787630796432495,
"step": 56
},
{
"epoch": 0.06863335340156532,
"grad_norm": 0.84517602222297,
"learning_rate": 6.666666666666667e-05,
"loss": 0.8722257018089294,
"step": 57
},
{
"epoch": 0.06983744732089103,
"grad_norm": 0.7596529781054532,
"learning_rate": 6.785714285714286e-05,
"loss": 0.8949298858642578,
"step": 58
},
{
"epoch": 0.07104154124021674,
"grad_norm": 0.6603054780063723,
"learning_rate": 6.904761904761905e-05,
"loss": 0.91631019115448,
"step": 59
},
{
"epoch": 0.07224563515954245,
"grad_norm": 0.6587245162237584,
"learning_rate": 7.023809523809524e-05,
"loss": 0.8479186296463013,
"step": 60
},
{
"epoch": 0.07344972907886815,
"grad_norm": 0.5896173965743856,
"learning_rate": 7.142857142857143e-05,
"loss": 0.8625457286834717,
"step": 61
},
{
"epoch": 0.07465382299819386,
"grad_norm": 0.656853432206036,
"learning_rate": 7.261904761904762e-05,
"loss": 0.7943159937858582,
"step": 62
},
{
"epoch": 0.07585791691751957,
"grad_norm": 0.7173445966189417,
"learning_rate": 7.380952380952382e-05,
"loss": 0.8757072687149048,
"step": 63
},
{
"epoch": 0.07706201083684527,
"grad_norm": 0.741123976756676,
"learning_rate": 7.500000000000001e-05,
"loss": 0.9337731003761292,
"step": 64
},
{
"epoch": 0.07826610475617098,
"grad_norm": 0.6587435716311675,
"learning_rate": 7.619047619047618e-05,
"loss": 0.9480957984924316,
"step": 65
},
{
"epoch": 0.07947019867549669,
"grad_norm": 0.7923118352415254,
"learning_rate": 7.738095238095239e-05,
"loss": 0.9424346089363098,
"step": 66
},
{
"epoch": 0.0806742925948224,
"grad_norm": 0.7437650621360334,
"learning_rate": 7.857142857142858e-05,
"loss": 0.9045975208282471,
"step": 67
},
{
"epoch": 0.08187838651414811,
"grad_norm": 0.7969199409769604,
"learning_rate": 7.976190476190477e-05,
"loss": 0.9725304245948792,
"step": 68
},
{
"epoch": 0.08308248043347381,
"grad_norm": 0.7206589585684046,
"learning_rate": 8.095238095238096e-05,
"loss": 0.9252752661705017,
"step": 69
},
{
"epoch": 0.08428657435279951,
"grad_norm": 0.8407014467727374,
"learning_rate": 8.214285714285714e-05,
"loss": 0.8761662244796753,
"step": 70
},
{
"epoch": 0.08549066827212523,
"grad_norm": 0.6884460334379702,
"learning_rate": 8.333333333333334e-05,
"loss": 0.8926453590393066,
"step": 71
},
{
"epoch": 0.08669476219145093,
"grad_norm": 0.7572421558690815,
"learning_rate": 8.452380952380952e-05,
"loss": 1.014685869216919,
"step": 72
},
{
"epoch": 0.08789885611077664,
"grad_norm": 0.7214896773064038,
"learning_rate": 8.571428571428571e-05,
"loss": 0.9112828969955444,
"step": 73
},
{
"epoch": 0.08910295003010235,
"grad_norm": 0.6504920809145635,
"learning_rate": 8.690476190476192e-05,
"loss": 0.877046525478363,
"step": 74
},
{
"epoch": 0.09030704394942805,
"grad_norm": 0.7130323126220806,
"learning_rate": 8.80952380952381e-05,
"loss": 0.988493025302887,
"step": 75
},
{
"epoch": 0.09151113786875377,
"grad_norm": 0.6759832506467568,
"learning_rate": 8.92857142857143e-05,
"loss": 0.8634282946586609,
"step": 76
},
{
"epoch": 0.09271523178807947,
"grad_norm": 0.7128524999186786,
"learning_rate": 9.047619047619048e-05,
"loss": 0.8843424320220947,
"step": 77
},
{
"epoch": 0.09391932570740517,
"grad_norm": 0.7527502094813464,
"learning_rate": 9.166666666666667e-05,
"loss": 0.9262052178382874,
"step": 78
},
{
"epoch": 0.09512341962673089,
"grad_norm": 0.679134529747516,
"learning_rate": 9.285714285714286e-05,
"loss": 0.902319610118866,
"step": 79
},
{
"epoch": 0.09632751354605659,
"grad_norm": 0.6582862889379956,
"learning_rate": 9.404761904761905e-05,
"loss": 0.9413686394691467,
"step": 80
},
{
"epoch": 0.0975316074653823,
"grad_norm": 0.6130866418518588,
"learning_rate": 9.523809523809524e-05,
"loss": 0.8581817150115967,
"step": 81
},
{
"epoch": 0.09873570138470801,
"grad_norm": 0.5687360865386506,
"learning_rate": 9.642857142857143e-05,
"loss": 0.873586893081665,
"step": 82
},
{
"epoch": 0.09993979530403371,
"grad_norm": 0.5975148331542051,
"learning_rate": 9.761904761904762e-05,
"loss": 0.8216943740844727,
"step": 83
},
{
"epoch": 0.10114388922335943,
"grad_norm": 0.5756637160773737,
"learning_rate": 9.880952380952381e-05,
"loss": 0.8205018639564514,
"step": 84
},
{
"epoch": 0.10234798314268513,
"grad_norm": 0.6452316625847548,
"learning_rate": 0.0001,
"loss": 0.9865813255310059,
"step": 85
},
{
"epoch": 0.10355207706201083,
"grad_norm": 0.6104006813254862,
"learning_rate": 9.999990091096183e-05,
"loss": 0.8462604284286499,
"step": 86
},
{
"epoch": 0.10475617098133655,
"grad_norm": 0.5499916520131942,
"learning_rate": 9.999960364424007e-05,
"loss": 0.8234490752220154,
"step": 87
},
{
"epoch": 0.10596026490066225,
"grad_norm": 0.5887689279983574,
"learning_rate": 9.999910820101293e-05,
"loss": 0.7696816921234131,
"step": 88
},
{
"epoch": 0.10716435881998795,
"grad_norm": 0.6295647282540509,
"learning_rate": 9.999841458324413e-05,
"loss": 0.9583300352096558,
"step": 89
},
{
"epoch": 0.10836845273931367,
"grad_norm": 0.6257819500551344,
"learning_rate": 9.999752279368291e-05,
"loss": 0.8089509010314941,
"step": 90
},
{
"epoch": 0.10957254665863937,
"grad_norm": 0.6853890534698811,
"learning_rate": 9.999643283586388e-05,
"loss": 0.7177246809005737,
"step": 91
},
{
"epoch": 0.11077664057796509,
"grad_norm": 0.5875425882208323,
"learning_rate": 9.999514471410719e-05,
"loss": 0.8235414028167725,
"step": 92
},
{
"epoch": 0.11198073449729079,
"grad_norm": 0.6657487103601663,
"learning_rate": 9.999365843351838e-05,
"loss": 0.878923773765564,
"step": 93
},
{
"epoch": 0.11318482841661649,
"grad_norm": 0.6312303761922631,
"learning_rate": 9.999197399998841e-05,
"loss": 0.9166843891143799,
"step": 94
},
{
"epoch": 0.11438892233594221,
"grad_norm": 0.6219532249650043,
"learning_rate": 9.999009142019364e-05,
"loss": 0.8822045922279358,
"step": 95
},
{
"epoch": 0.11559301625526791,
"grad_norm": 0.6714545958652653,
"learning_rate": 9.99880107015958e-05,
"loss": 0.8442031145095825,
"step": 96
},
{
"epoch": 0.11679711017459361,
"grad_norm": 0.6187610552353822,
"learning_rate": 9.998573185244192e-05,
"loss": 0.7852096557617188,
"step": 97
},
{
"epoch": 0.11800120409391933,
"grad_norm": 0.578079618701512,
"learning_rate": 9.998325488176437e-05,
"loss": 0.8714963793754578,
"step": 98
},
{
"epoch": 0.11920529801324503,
"grad_norm": 0.5345630777945766,
"learning_rate": 9.99805797993808e-05,
"loss": 0.8692537546157837,
"step": 99
},
{
"epoch": 0.12040939193257075,
"grad_norm": 0.6181542688226837,
"learning_rate": 9.997770661589403e-05,
"loss": 0.8973393440246582,
"step": 100
},
{
"epoch": 0.12161348585189645,
"grad_norm": 0.64002745884027,
"learning_rate": 9.997463534269213e-05,
"loss": 0.8967085480690002,
"step": 101
},
{
"epoch": 0.12281757977122215,
"grad_norm": 0.7341991040389423,
"learning_rate": 9.997136599194825e-05,
"loss": 0.8160005807876587,
"step": 102
},
{
"epoch": 0.12402167369054787,
"grad_norm": 0.6265972559860278,
"learning_rate": 9.996789857662068e-05,
"loss": 1.0133693218231201,
"step": 103
},
{
"epoch": 0.12522576760987358,
"grad_norm": 0.6957185688667177,
"learning_rate": 9.996423311045273e-05,
"loss": 0.8984023928642273,
"step": 104
},
{
"epoch": 0.12642986152919927,
"grad_norm": 0.5956498399306159,
"learning_rate": 9.99603696079727e-05,
"loss": 0.9195867776870728,
"step": 105
},
{
"epoch": 0.127633955448525,
"grad_norm": 0.6463966531285172,
"learning_rate": 9.995630808449383e-05,
"loss": 0.8191015720367432,
"step": 106
},
{
"epoch": 0.1288380493678507,
"grad_norm": 0.6171569033866292,
"learning_rate": 9.99520485561142e-05,
"loss": 0.9108123779296875,
"step": 107
},
{
"epoch": 0.1300421432871764,
"grad_norm": 0.6143974868615907,
"learning_rate": 9.994759103971673e-05,
"loss": 0.8896417617797852,
"step": 108
},
{
"epoch": 0.1312462372065021,
"grad_norm": 0.6149896368980597,
"learning_rate": 9.994293555296904e-05,
"loss": 0.8559168577194214,
"step": 109
},
{
"epoch": 0.13245033112582782,
"grad_norm": 0.6515351535933014,
"learning_rate": 9.993808211432346e-05,
"loss": 0.8588761687278748,
"step": 110
},
{
"epoch": 0.1336544250451535,
"grad_norm": 0.6676952442415878,
"learning_rate": 9.993303074301689e-05,
"loss": 0.8163575530052185,
"step": 111
},
{
"epoch": 0.13485851896447923,
"grad_norm": 0.6375806365685596,
"learning_rate": 9.992778145907073e-05,
"loss": 0.9100340604782104,
"step": 112
},
{
"epoch": 0.13606261288380495,
"grad_norm": 0.6252693886741836,
"learning_rate": 9.992233428329088e-05,
"loss": 0.8442620038986206,
"step": 113
},
{
"epoch": 0.13726670680313063,
"grad_norm": 0.7244683282798454,
"learning_rate": 9.99166892372675e-05,
"loss": 0.8886062502861023,
"step": 114
},
{
"epoch": 0.13847080072245635,
"grad_norm": 0.7464382496737153,
"learning_rate": 9.991084634337511e-05,
"loss": 0.8216350674629211,
"step": 115
},
{
"epoch": 0.13967489464178207,
"grad_norm": 0.646628205480326,
"learning_rate": 9.99048056247724e-05,
"loss": 0.9128209948539734,
"step": 116
},
{
"epoch": 0.14087898856110775,
"grad_norm": 0.5712623706680054,
"learning_rate": 9.989856710540209e-05,
"loss": 0.8703181743621826,
"step": 117
},
{
"epoch": 0.14208308248043347,
"grad_norm": 0.6329274208544833,
"learning_rate": 9.989213080999097e-05,
"loss": 0.8345284461975098,
"step": 118
},
{
"epoch": 0.1432871763997592,
"grad_norm": 0.62852450343761,
"learning_rate": 9.988549676404965e-05,
"loss": 0.8411127328872681,
"step": 119
},
{
"epoch": 0.1444912703190849,
"grad_norm": 0.6420543736879645,
"learning_rate": 9.987866499387264e-05,
"loss": 0.9257729649543762,
"step": 120
},
{
"epoch": 0.1456953642384106,
"grad_norm": 0.5528124716336356,
"learning_rate": 9.987163552653802e-05,
"loss": 0.8293143510818481,
"step": 121
},
{
"epoch": 0.1468994581577363,
"grad_norm": 0.6033234224852082,
"learning_rate": 9.986440838990755e-05,
"loss": 0.8318862915039062,
"step": 122
},
{
"epoch": 0.14810355207706202,
"grad_norm": 0.5677902467857496,
"learning_rate": 9.985698361262642e-05,
"loss": 0.8512689471244812,
"step": 123
},
{
"epoch": 0.1493076459963877,
"grad_norm": 0.6052037467329195,
"learning_rate": 9.984936122412319e-05,
"loss": 0.8174145221710205,
"step": 124
},
{
"epoch": 0.15051173991571343,
"grad_norm": 0.6203272750504745,
"learning_rate": 9.984154125460969e-05,
"loss": 0.8040434718132019,
"step": 125
},
{
"epoch": 0.15171583383503914,
"grad_norm": 0.5640742447104545,
"learning_rate": 9.983352373508081e-05,
"loss": 0.8144879341125488,
"step": 126
},
{
"epoch": 0.15291992775436483,
"grad_norm": 0.7383291925432525,
"learning_rate": 9.982530869731451e-05,
"loss": 0.8918353915214539,
"step": 127
},
{
"epoch": 0.15412402167369055,
"grad_norm": 0.6110098020804609,
"learning_rate": 9.98168961738716e-05,
"loss": 0.8100703358650208,
"step": 128
},
{
"epoch": 0.15532811559301626,
"grad_norm": 0.7276368732153812,
"learning_rate": 9.980828619809561e-05,
"loss": 0.8626886010169983,
"step": 129
},
{
"epoch": 0.15653220951234195,
"grad_norm": 0.6182775041438267,
"learning_rate": 9.979947880411273e-05,
"loss": 0.7800024747848511,
"step": 130
},
{
"epoch": 0.15773630343166767,
"grad_norm": 0.6684016472398361,
"learning_rate": 9.97904740268316e-05,
"loss": 0.8336461782455444,
"step": 131
},
{
"epoch": 0.15894039735099338,
"grad_norm": 0.6781202957252627,
"learning_rate": 9.978127190194322e-05,
"loss": 0.9167442321777344,
"step": 132
},
{
"epoch": 0.16014449127031907,
"grad_norm": 0.5781488999510032,
"learning_rate": 9.977187246592076e-05,
"loss": 0.8317021727561951,
"step": 133
},
{
"epoch": 0.1613485851896448,
"grad_norm": 0.558442522325844,
"learning_rate": 9.976227575601947e-05,
"loss": 0.7598085999488831,
"step": 134
},
{
"epoch": 0.1625526791089705,
"grad_norm": 0.641636921915755,
"learning_rate": 9.97524818102765e-05,
"loss": 0.837604284286499,
"step": 135
},
{
"epoch": 0.16375677302829622,
"grad_norm": 0.6091929337726231,
"learning_rate": 9.974249066751077e-05,
"loss": 0.9500613212585449,
"step": 136
},
{
"epoch": 0.1649608669476219,
"grad_norm": 0.6466178033726994,
"learning_rate": 9.973230236732276e-05,
"loss": 0.771168053150177,
"step": 137
},
{
"epoch": 0.16616496086694763,
"grad_norm": 0.6660733398000859,
"learning_rate": 9.972191695009446e-05,
"loss": 0.7790799140930176,
"step": 138
},
{
"epoch": 0.16736905478627334,
"grad_norm": 0.6115980204086684,
"learning_rate": 9.971133445698908e-05,
"loss": 0.7652186155319214,
"step": 139
},
{
"epoch": 0.16857314870559903,
"grad_norm": 0.5413962093401579,
"learning_rate": 9.9700554929951e-05,
"loss": 0.7228227257728577,
"step": 140
},
{
"epoch": 0.16977724262492475,
"grad_norm": 0.5939091006500794,
"learning_rate": 9.968957841170554e-05,
"loss": 0.8524500131607056,
"step": 141
},
{
"epoch": 0.17098133654425046,
"grad_norm": 0.6504939461933902,
"learning_rate": 9.967840494575879e-05,
"loss": 0.887976348400116,
"step": 142
},
{
"epoch": 0.17218543046357615,
"grad_norm": 0.6430414793958295,
"learning_rate": 9.966703457639748e-05,
"loss": 0.880845308303833,
"step": 143
},
{
"epoch": 0.17338952438290187,
"grad_norm": 0.5968094588570781,
"learning_rate": 9.965546734868875e-05,
"loss": 0.7715303897857666,
"step": 144
},
{
"epoch": 0.17459361830222758,
"grad_norm": 0.6214316919780661,
"learning_rate": 9.964370330848005e-05,
"loss": 0.7617921829223633,
"step": 145
},
{
"epoch": 0.17579771222155327,
"grad_norm": 0.6945966190350599,
"learning_rate": 9.963174250239888e-05,
"loss": 0.8942645788192749,
"step": 146
},
{
"epoch": 0.177001806140879,
"grad_norm": 0.5946158752732297,
"learning_rate": 9.96195849778526e-05,
"loss": 0.8035511374473572,
"step": 147
},
{
"epoch": 0.1782059000602047,
"grad_norm": 0.5570594020442262,
"learning_rate": 9.960723078302832e-05,
"loss": 0.746635377407074,
"step": 148
},
{
"epoch": 0.1794099939795304,
"grad_norm": 0.5300767685887228,
"learning_rate": 9.959467996689264e-05,
"loss": 0.788428783416748,
"step": 149
},
{
"epoch": 0.1806140878988561,
"grad_norm": 0.6429771066659856,
"learning_rate": 9.95819325791915e-05,
"loss": 0.8748104572296143,
"step": 150
},
{
"epoch": 0.18181818181818182,
"grad_norm": 0.6009681075436003,
"learning_rate": 9.956898867044999e-05,
"loss": 0.8715510368347168,
"step": 151
},
{
"epoch": 0.18302227573750754,
"grad_norm": 0.6287471875141659,
"learning_rate": 9.955584829197203e-05,
"loss": 0.9487425088882446,
"step": 152
},
{
"epoch": 0.18422636965683323,
"grad_norm": 0.5832830587585199,
"learning_rate": 9.954251149584036e-05,
"loss": 0.7733198404312134,
"step": 153
},
{
"epoch": 0.18543046357615894,
"grad_norm": 0.5985330987653703,
"learning_rate": 9.952897833491617e-05,
"loss": 0.8523417711257935,
"step": 154
},
{
"epoch": 0.18663455749548466,
"grad_norm": 0.5714807380926865,
"learning_rate": 9.951524886283899e-05,
"loss": 0.7397559881210327,
"step": 155
},
{
"epoch": 0.18783865141481035,
"grad_norm": 0.6440313958900563,
"learning_rate": 9.950132313402642e-05,
"loss": 0.8231230974197388,
"step": 156
},
{
"epoch": 0.18904274533413606,
"grad_norm": 0.7067535788120153,
"learning_rate": 9.948720120367394e-05,
"loss": 0.8909597396850586,
"step": 157
},
{
"epoch": 0.19024683925346178,
"grad_norm": 0.6005944043833608,
"learning_rate": 9.947288312775471e-05,
"loss": 0.7262487411499023,
"step": 158
},
{
"epoch": 0.19145093317278747,
"grad_norm": 0.6029984307683901,
"learning_rate": 9.945836896301927e-05,
"loss": 0.8319439888000488,
"step": 159
},
{
"epoch": 0.19265502709211318,
"grad_norm": 0.6224876412877436,
"learning_rate": 9.944365876699544e-05,
"loss": 0.8430992364883423,
"step": 160
},
{
"epoch": 0.1938591210114389,
"grad_norm": 0.6160811180626786,
"learning_rate": 9.942875259798796e-05,
"loss": 0.7832828760147095,
"step": 161
},
{
"epoch": 0.1950632149307646,
"grad_norm": 0.6218195815569507,
"learning_rate": 9.941365051507836e-05,
"loss": 0.9227669835090637,
"step": 162
},
{
"epoch": 0.1962673088500903,
"grad_norm": 0.5986059436334881,
"learning_rate": 9.939835257812468e-05,
"loss": 0.9111921787261963,
"step": 163
},
{
"epoch": 0.19747140276941602,
"grad_norm": 0.5881408038735889,
"learning_rate": 9.938285884776121e-05,
"loss": 0.8965997695922852,
"step": 164
},
{
"epoch": 0.1986754966887417,
"grad_norm": 0.6374823214972137,
"learning_rate": 9.936716938539834e-05,
"loss": 0.8731657266616821,
"step": 165
},
{
"epoch": 0.19987959060806743,
"grad_norm": 0.6161832126609788,
"learning_rate": 9.93512842532222e-05,
"loss": 0.8580436706542969,
"step": 166
},
{
"epoch": 0.20108368452739314,
"grad_norm": 0.553470250749396,
"learning_rate": 9.93352035141945e-05,
"loss": 0.7036675810813904,
"step": 167
},
{
"epoch": 0.20228777844671886,
"grad_norm": 0.7770006457997596,
"learning_rate": 9.931892723205221e-05,
"loss": 0.8978402614593506,
"step": 168
},
{
"epoch": 0.20349187236604455,
"grad_norm": 0.5743393143744865,
"learning_rate": 9.93024554713074e-05,
"loss": 0.7655681371688843,
"step": 169
},
{
"epoch": 0.20469596628537026,
"grad_norm": 0.5517387700265929,
"learning_rate": 9.92857882972469e-05,
"loss": 0.797248899936676,
"step": 170
},
{
"epoch": 0.20590006020469598,
"grad_norm": 0.5547295397097244,
"learning_rate": 9.926892577593208e-05,
"loss": 0.8540539741516113,
"step": 171
},
{
"epoch": 0.20710415412402167,
"grad_norm": 0.5964203337683545,
"learning_rate": 9.925186797419858e-05,
"loss": 0.8034850358963013,
"step": 172
},
{
"epoch": 0.20830824804334738,
"grad_norm": 0.6189183804758942,
"learning_rate": 9.923461495965607e-05,
"loss": 0.8679941892623901,
"step": 173
},
{
"epoch": 0.2095123419626731,
"grad_norm": 0.5866776954815867,
"learning_rate": 9.92171668006879e-05,
"loss": 0.7853211164474487,
"step": 174
},
{
"epoch": 0.2107164358819988,
"grad_norm": 0.5643948583479728,
"learning_rate": 9.919952356645092e-05,
"loss": 0.8060238361358643,
"step": 175
},
{
"epoch": 0.2119205298013245,
"grad_norm": 0.5823524489669061,
"learning_rate": 9.91816853268752e-05,
"loss": 0.8626150488853455,
"step": 176
},
{
"epoch": 0.21312462372065022,
"grad_norm": 0.5622075738754846,
"learning_rate": 9.91636521526637e-05,
"loss": 0.8187872767448425,
"step": 177
},
{
"epoch": 0.2143287176399759,
"grad_norm": 0.6733984895343411,
"learning_rate": 9.9145424115292e-05,
"loss": 0.8441307544708252,
"step": 178
},
{
"epoch": 0.21553281155930162,
"grad_norm": 0.5333453551939791,
"learning_rate": 9.912700128700805e-05,
"loss": 0.7993243932723999,
"step": 179
},
{
"epoch": 0.21673690547862734,
"grad_norm": 0.555877229699453,
"learning_rate": 9.910838374083185e-05,
"loss": 0.774019181728363,
"step": 180
},
{
"epoch": 0.21794099939795303,
"grad_norm": 0.6391151018213997,
"learning_rate": 9.908957155055523e-05,
"loss": 0.8688267469406128,
"step": 181
},
{
"epoch": 0.21914509331727874,
"grad_norm": 0.6241625884954857,
"learning_rate": 9.907056479074142e-05,
"loss": 0.847202718257904,
"step": 182
},
{
"epoch": 0.22034918723660446,
"grad_norm": 0.5890469266154632,
"learning_rate": 9.905136353672492e-05,
"loss": 0.7770888209342957,
"step": 183
},
{
"epoch": 0.22155328115593018,
"grad_norm": 0.5769999378849722,
"learning_rate": 9.903196786461106e-05,
"loss": 0.7439048290252686,
"step": 184
},
{
"epoch": 0.22275737507525586,
"grad_norm": 0.5754784178675677,
"learning_rate": 9.901237785127577e-05,
"loss": 0.8247470855712891,
"step": 185
},
{
"epoch": 0.22396146899458158,
"grad_norm": 0.5933793451997219,
"learning_rate": 9.89925935743653e-05,
"loss": 0.8449085354804993,
"step": 186
},
{
"epoch": 0.2251655629139073,
"grad_norm": 0.6188199832931084,
"learning_rate": 9.897261511229583e-05,
"loss": 0.9046580195426941,
"step": 187
},
{
"epoch": 0.22636965683323299,
"grad_norm": 0.5333245358703447,
"learning_rate": 9.895244254425322e-05,
"loss": 0.7648054957389832,
"step": 188
},
{
"epoch": 0.2275737507525587,
"grad_norm": 0.6143930481556643,
"learning_rate": 9.893207595019269e-05,
"loss": 0.8550212383270264,
"step": 189
},
{
"epoch": 0.22877784467188442,
"grad_norm": 0.5796146274953775,
"learning_rate": 9.891151541083852e-05,
"loss": 0.8073182702064514,
"step": 190
},
{
"epoch": 0.2299819385912101,
"grad_norm": 0.6327168339480935,
"learning_rate": 9.889076100768362e-05,
"loss": 0.8802028298377991,
"step": 191
},
{
"epoch": 0.23118603251053582,
"grad_norm": 0.6340397780374257,
"learning_rate": 9.886981282298937e-05,
"loss": 0.8678988218307495,
"step": 192
},
{
"epoch": 0.23239012642986154,
"grad_norm": 0.5781580859543161,
"learning_rate": 9.884867093978519e-05,
"loss": 0.7214905023574829,
"step": 193
},
{
"epoch": 0.23359422034918723,
"grad_norm": 0.5864889560079226,
"learning_rate": 9.882733544186825e-05,
"loss": 0.8775925636291504,
"step": 194
},
{
"epoch": 0.23479831426851294,
"grad_norm": 0.6195540628078187,
"learning_rate": 9.880580641380306e-05,
"loss": 0.7874777317047119,
"step": 195
},
{
"epoch": 0.23600240818783866,
"grad_norm": 0.6567543350650362,
"learning_rate": 9.878408394092129e-05,
"loss": 0.8925746083259583,
"step": 196
},
{
"epoch": 0.23720650210716435,
"grad_norm": 0.6610048982837163,
"learning_rate": 9.876216810932129e-05,
"loss": 0.8471571207046509,
"step": 197
},
{
"epoch": 0.23841059602649006,
"grad_norm": 0.6868969238880852,
"learning_rate": 9.87400590058678e-05,
"loss": 0.8910620808601379,
"step": 198
},
{
"epoch": 0.23961468994581578,
"grad_norm": 0.6370290531097156,
"learning_rate": 9.871775671819162e-05,
"loss": 0.8550009727478027,
"step": 199
},
{
"epoch": 0.2408187838651415,
"grad_norm": 0.5905145265636871,
"learning_rate": 9.869526133468923e-05,
"loss": 0.7858701348304749,
"step": 200
},
{
"epoch": 0.24202287778446718,
"grad_norm": 0.5396495024692038,
"learning_rate": 9.867257294452245e-05,
"loss": 0.7874826192855835,
"step": 201
},
{
"epoch": 0.2432269717037929,
"grad_norm": 0.5611619691570671,
"learning_rate": 9.864969163761817e-05,
"loss": 0.8445602059364319,
"step": 202
},
{
"epoch": 0.24443106562311862,
"grad_norm": 0.5927387531343714,
"learning_rate": 9.86266175046678e-05,
"loss": 0.7110156416893005,
"step": 203
},
{
"epoch": 0.2456351595424443,
"grad_norm": 0.5667132102728043,
"learning_rate": 9.86033506371271e-05,
"loss": 0.7977332472801208,
"step": 204
},
{
"epoch": 0.24683925346177002,
"grad_norm": 0.6083484012142651,
"learning_rate": 9.857989112721574e-05,
"loss": 0.7608157992362976,
"step": 205
},
{
"epoch": 0.24804334738109574,
"grad_norm": 0.5943891087363417,
"learning_rate": 9.855623906791693e-05,
"loss": 0.7913424968719482,
"step": 206
},
{
"epoch": 0.24924744130042142,
"grad_norm": 0.5548488707009894,
"learning_rate": 9.853239455297705e-05,
"loss": 0.851811945438385,
"step": 207
},
{
"epoch": 0.25045153521974717,
"grad_norm": 0.5095216387061532,
"learning_rate": 9.850835767690532e-05,
"loss": 0.7788999676704407,
"step": 208
},
{
"epoch": 0.25165562913907286,
"grad_norm": 0.6581296320927581,
"learning_rate": 9.848412853497337e-05,
"loss": 0.852281391620636,
"step": 209
},
{
"epoch": 0.25285972305839854,
"grad_norm": 0.5624078228873652,
"learning_rate": 9.845970722321489e-05,
"loss": 0.7634679079055786,
"step": 210
},
{
"epoch": 0.2540638169777243,
"grad_norm": 0.58279073197933,
"learning_rate": 9.843509383842525e-05,
"loss": 0.8002668619155884,
"step": 211
},
{
"epoch": 0.25526791089705,
"grad_norm": 0.602131782992255,
"learning_rate": 9.841028847816112e-05,
"loss": 0.917669951915741,
"step": 212
},
{
"epoch": 0.25647200481637566,
"grad_norm": 0.5824275546138442,
"learning_rate": 9.838529124074006e-05,
"loss": 0.6941547393798828,
"step": 213
},
{
"epoch": 0.2576760987357014,
"grad_norm": 0.6855988184994416,
"learning_rate": 9.836010222524018e-05,
"loss": 0.7900656461715698,
"step": 214
},
{
"epoch": 0.2588801926550271,
"grad_norm": 0.6033793243686978,
"learning_rate": 9.833472153149968e-05,
"loss": 0.7909616231918335,
"step": 215
},
{
"epoch": 0.2600842865743528,
"grad_norm": 0.5558985979936094,
"learning_rate": 9.830914926011651e-05,
"loss": 0.8461582660675049,
"step": 216
},
{
"epoch": 0.26128838049367853,
"grad_norm": 0.6135755856519033,
"learning_rate": 9.828338551244794e-05,
"loss": 0.8185094594955444,
"step": 217
},
{
"epoch": 0.2624924744130042,
"grad_norm": 0.5913147304313477,
"learning_rate": 9.825743039061015e-05,
"loss": 0.7694218754768372,
"step": 218
},
{
"epoch": 0.2636965683323299,
"grad_norm": 0.5826277813793379,
"learning_rate": 9.823128399747788e-05,
"loss": 0.9048447012901306,
"step": 219
},
{
"epoch": 0.26490066225165565,
"grad_norm": 0.6069752354101805,
"learning_rate": 9.820494643668396e-05,
"loss": 0.7263599038124084,
"step": 220
},
{
"epoch": 0.26610475617098134,
"grad_norm": 0.5779260861618932,
"learning_rate": 9.817841781261894e-05,
"loss": 0.7507243156433105,
"step": 221
},
{
"epoch": 0.267308850090307,
"grad_norm": 0.5407800371575219,
"learning_rate": 9.815169823043066e-05,
"loss": 0.7708674073219299,
"step": 222
},
{
"epoch": 0.26851294400963277,
"grad_norm": 0.5354161457972056,
"learning_rate": 9.812478779602381e-05,
"loss": 0.8195161819458008,
"step": 223
},
{
"epoch": 0.26971703792895846,
"grad_norm": 0.5435156479444293,
"learning_rate": 9.809768661605956e-05,
"loss": 0.8014235496520996,
"step": 224
},
{
"epoch": 0.27092113184828415,
"grad_norm": 0.6451511656258695,
"learning_rate": 9.80703947979551e-05,
"loss": 0.7672905921936035,
"step": 225
},
{
"epoch": 0.2721252257676099,
"grad_norm": 0.5770638130924962,
"learning_rate": 9.804291244988324e-05,
"loss": 0.8396711349487305,
"step": 226
},
{
"epoch": 0.2733293196869356,
"grad_norm": 0.6047093718584222,
"learning_rate": 9.801523968077195e-05,
"loss": 0.8380450010299683,
"step": 227
},
{
"epoch": 0.27453341360626127,
"grad_norm": 0.5925957423670997,
"learning_rate": 9.798737660030397e-05,
"loss": 0.8844544887542725,
"step": 228
},
{
"epoch": 0.275737507525587,
"grad_norm": 0.5982585104488206,
"learning_rate": 9.79593233189163e-05,
"loss": 0.9639543294906616,
"step": 229
},
{
"epoch": 0.2769416014449127,
"grad_norm": 0.5851751603385734,
"learning_rate": 9.793107994779988e-05,
"loss": 0.7578297853469849,
"step": 230
},
{
"epoch": 0.2781456953642384,
"grad_norm": 0.5662418714139069,
"learning_rate": 9.790264659889903e-05,
"loss": 0.7516130208969116,
"step": 231
},
{
"epoch": 0.27934978928356413,
"grad_norm": 0.5629010447957682,
"learning_rate": 9.787402338491108e-05,
"loss": 0.813753604888916,
"step": 232
},
{
"epoch": 0.2805538832028898,
"grad_norm": 0.611997046314177,
"learning_rate": 9.78452104192859e-05,
"loss": 0.7318404912948608,
"step": 233
},
{
"epoch": 0.2817579771222155,
"grad_norm": 0.6083078236956858,
"learning_rate": 9.781620781622546e-05,
"loss": 0.7795756459236145,
"step": 234
},
{
"epoch": 0.28296207104154125,
"grad_norm": 0.5733435457346814,
"learning_rate": 9.778701569068336e-05,
"loss": 0.7374966740608215,
"step": 235
},
{
"epoch": 0.28416616496086694,
"grad_norm": 0.5854632993757632,
"learning_rate": 9.775763415836439e-05,
"loss": 0.8152198791503906,
"step": 236
},
{
"epoch": 0.28537025888019263,
"grad_norm": 0.5754299346488221,
"learning_rate": 9.772806333572405e-05,
"loss": 0.8636054992675781,
"step": 237
},
{
"epoch": 0.2865743527995184,
"grad_norm": 0.5422881955560872,
"learning_rate": 9.76983033399681e-05,
"loss": 0.7844322919845581,
"step": 238
},
{
"epoch": 0.28777844671884406,
"grad_norm": 0.6194774799956613,
"learning_rate": 9.766835428905213e-05,
"loss": 0.7630101442337036,
"step": 239
},
{
"epoch": 0.2889825406381698,
"grad_norm": 0.603975682522893,
"learning_rate": 9.763821630168106e-05,
"loss": 0.7873789668083191,
"step": 240
},
{
"epoch": 0.2901866345574955,
"grad_norm": 0.6133591002499805,
"learning_rate": 9.760788949730866e-05,
"loss": 0.8574675917625427,
"step": 241
},
{
"epoch": 0.2913907284768212,
"grad_norm": 0.5274827858620643,
"learning_rate": 9.757737399613706e-05,
"loss": 0.7098798751831055,
"step": 242
},
{
"epoch": 0.2925948223961469,
"grad_norm": 0.5891469808235313,
"learning_rate": 9.754666991911633e-05,
"loss": 0.8549659252166748,
"step": 243
},
{
"epoch": 0.2937989163154726,
"grad_norm": 0.6778037762349854,
"learning_rate": 9.751577738794398e-05,
"loss": 0.8683890700340271,
"step": 244
},
{
"epoch": 0.2950030102347983,
"grad_norm": 0.5529589488208931,
"learning_rate": 9.748469652506446e-05,
"loss": 0.8280667066574097,
"step": 245
},
{
"epoch": 0.29620710415412405,
"grad_norm": 0.6019791904555852,
"learning_rate": 9.745342745366867e-05,
"loss": 0.8362815380096436,
"step": 246
},
{
"epoch": 0.29741119807344973,
"grad_norm": 0.6103921005737301,
"learning_rate": 9.74219702976935e-05,
"loss": 0.7211261987686157,
"step": 247
},
{
"epoch": 0.2986152919927754,
"grad_norm": 0.5250093329310034,
"learning_rate": 9.739032518182134e-05,
"loss": 0.8164722919464111,
"step": 248
},
{
"epoch": 0.29981938591210117,
"grad_norm": 0.5388495396934913,
"learning_rate": 9.735849223147952e-05,
"loss": 0.7128286361694336,
"step": 249
},
{
"epoch": 0.30102347983142685,
"grad_norm": 0.5829439265945735,
"learning_rate": 9.732647157283994e-05,
"loss": 0.8446158766746521,
"step": 250
},
{
"epoch": 0.30222757375075254,
"grad_norm": 0.5295167319429783,
"learning_rate": 9.729426333281841e-05,
"loss": 0.7621672749519348,
"step": 251
},
{
"epoch": 0.3034316676700783,
"grad_norm": 0.8911053202384595,
"learning_rate": 9.726186763907432e-05,
"loss": 0.8294147253036499,
"step": 252
},
{
"epoch": 0.304635761589404,
"grad_norm": 0.5529587210931857,
"learning_rate": 9.722928462000995e-05,
"loss": 0.7578885555267334,
"step": 253
},
{
"epoch": 0.30583985550872966,
"grad_norm": 0.6002485806157559,
"learning_rate": 9.71965144047701e-05,
"loss": 0.7943164110183716,
"step": 254
},
{
"epoch": 0.3070439494280554,
"grad_norm": 0.5694226586132322,
"learning_rate": 9.716355712324156e-05,
"loss": 0.8210703134536743,
"step": 255
},
{
"epoch": 0.3082480433473811,
"grad_norm": 0.6922498599198393,
"learning_rate": 9.713041290605254e-05,
"loss": 0.873950719833374,
"step": 256
},
{
"epoch": 0.3094521372667068,
"grad_norm": 0.5445472317700759,
"learning_rate": 9.709708188457218e-05,
"loss": 0.8812997341156006,
"step": 257
},
{
"epoch": 0.3106562311860325,
"grad_norm": 0.5722532810838292,
"learning_rate": 9.706356419091003e-05,
"loss": 0.854279637336731,
"step": 258
},
{
"epoch": 0.3118603251053582,
"grad_norm": 0.6050281714134376,
"learning_rate": 9.702985995791554e-05,
"loss": 0.7731985449790955,
"step": 259
},
{
"epoch": 0.3130644190246839,
"grad_norm": 0.5259270771468094,
"learning_rate": 9.69959693191775e-05,
"loss": 0.7495138645172119,
"step": 260
},
{
"epoch": 0.31426851294400965,
"grad_norm": 0.599702214924439,
"learning_rate": 9.696189240902353e-05,
"loss": 0.8307251334190369,
"step": 261
},
{
"epoch": 0.31547260686333534,
"grad_norm": 0.5772891113584918,
"learning_rate": 9.69276293625196e-05,
"loss": 0.7606898546218872,
"step": 262
},
{
"epoch": 0.316676700782661,
"grad_norm": 0.6048200131150295,
"learning_rate": 9.689318031546935e-05,
"loss": 0.7834568023681641,
"step": 263
},
{
"epoch": 0.31788079470198677,
"grad_norm": 0.5187602167429113,
"learning_rate": 9.685854540441374e-05,
"loss": 0.8568704724311829,
"step": 264
},
{
"epoch": 0.31908488862131246,
"grad_norm": 0.589371193118731,
"learning_rate": 9.682372476663037e-05,
"loss": 0.8562246561050415,
"step": 265
},
{
"epoch": 0.32028898254063815,
"grad_norm": 0.578899694521127,
"learning_rate": 9.678871854013296e-05,
"loss": 0.7672446966171265,
"step": 266
},
{
"epoch": 0.3214930764599639,
"grad_norm": 0.6249280291756892,
"learning_rate": 9.675352686367086e-05,
"loss": 0.7922170162200928,
"step": 267
},
{
"epoch": 0.3226971703792896,
"grad_norm": 0.5749398173798772,
"learning_rate": 9.671814987672842e-05,
"loss": 0.8341658115386963,
"step": 268
},
{
"epoch": 0.32390126429861527,
"grad_norm": 0.6275533667804223,
"learning_rate": 9.668258771952453e-05,
"loss": 0.807477593421936,
"step": 269
},
{
"epoch": 0.325105358217941,
"grad_norm": 0.5821581899182365,
"learning_rate": 9.664684053301199e-05,
"loss": 0.7355650663375854,
"step": 270
},
{
"epoch": 0.3263094521372667,
"grad_norm": 0.5858441049798361,
"learning_rate": 9.661090845887693e-05,
"loss": 0.870678186416626,
"step": 271
},
{
"epoch": 0.32751354605659244,
"grad_norm": 0.5365325287676738,
"learning_rate": 9.657479163953839e-05,
"loss": 0.8608752489089966,
"step": 272
},
{
"epoch": 0.32871763997591813,
"grad_norm": 0.5599944247690204,
"learning_rate": 9.653849021814759e-05,
"loss": 0.8878623247146606,
"step": 273
},
{
"epoch": 0.3299217338952438,
"grad_norm": 0.5563052433880087,
"learning_rate": 9.650200433858741e-05,
"loss": 0.8141953945159912,
"step": 274
},
{
"epoch": 0.33112582781456956,
"grad_norm": 0.5535164304716184,
"learning_rate": 9.646533414547193e-05,
"loss": 0.7778948545455933,
"step": 275
},
{
"epoch": 0.33232992173389525,
"grad_norm": 0.6014215940932262,
"learning_rate": 9.642847978414569e-05,
"loss": 0.7866028547286987,
"step": 276
},
{
"epoch": 0.33353401565322094,
"grad_norm": 0.5331079808996885,
"learning_rate": 9.639144140068324e-05,
"loss": 0.7266688346862793,
"step": 277
},
{
"epoch": 0.3347381095725467,
"grad_norm": 0.5766611369648224,
"learning_rate": 9.635421914188845e-05,
"loss": 0.8532525300979614,
"step": 278
},
{
"epoch": 0.33594220349187237,
"grad_norm": 0.5358777135832988,
"learning_rate": 9.631681315529408e-05,
"loss": 0.8211830854415894,
"step": 279
},
{
"epoch": 0.33714629741119806,
"grad_norm": 0.6298604183968022,
"learning_rate": 9.627922358916102e-05,
"loss": 0.8692326545715332,
"step": 280
},
{
"epoch": 0.3383503913305238,
"grad_norm": 0.5797251655827115,
"learning_rate": 9.624145059247787e-05,
"loss": 0.8560823798179626,
"step": 281
},
{
"epoch": 0.3395544852498495,
"grad_norm": 0.5434794437969054,
"learning_rate": 9.620349431496018e-05,
"loss": 0.8305436968803406,
"step": 282
},
{
"epoch": 0.3407585791691752,
"grad_norm": 0.6441021904672944,
"learning_rate": 9.616535490705004e-05,
"loss": 0.7797948122024536,
"step": 283
},
{
"epoch": 0.3419626730885009,
"grad_norm": 0.5598221419768312,
"learning_rate": 9.612703251991528e-05,
"loss": 0.7669990062713623,
"step": 284
},
{
"epoch": 0.3431667670078266,
"grad_norm": 0.5580548895181614,
"learning_rate": 9.60885273054491e-05,
"loss": 0.8091415166854858,
"step": 285
},
{
"epoch": 0.3443708609271523,
"grad_norm": 0.6911896360916333,
"learning_rate": 9.604983941626924e-05,
"loss": 0.8437607288360596,
"step": 286
},
{
"epoch": 0.34557495484647804,
"grad_norm": 0.6177635146844122,
"learning_rate": 9.601096900571757e-05,
"loss": 0.891084611415863,
"step": 287
},
{
"epoch": 0.34677904876580373,
"grad_norm": 0.5777148183335996,
"learning_rate": 9.597191622785932e-05,
"loss": 0.8502274751663208,
"step": 288
},
{
"epoch": 0.3479831426851294,
"grad_norm": 0.5809974008489335,
"learning_rate": 9.593268123748259e-05,
"loss": 0.8225856423377991,
"step": 289
},
{
"epoch": 0.34918723660445516,
"grad_norm": 0.6051633632643184,
"learning_rate": 9.589326419009768e-05,
"loss": 0.7742931842803955,
"step": 290
},
{
"epoch": 0.35039133052378085,
"grad_norm": 0.5650358876193098,
"learning_rate": 9.585366524193648e-05,
"loss": 0.7171435356140137,
"step": 291
},
{
"epoch": 0.35159542444310654,
"grad_norm": 0.4774910718398751,
"learning_rate": 9.581388454995187e-05,
"loss": 0.6769593954086304,
"step": 292
},
{
"epoch": 0.3527995183624323,
"grad_norm": 0.6156416845975364,
"learning_rate": 9.577392227181706e-05,
"loss": 0.7639175653457642,
"step": 293
},
{
"epoch": 0.354003612281758,
"grad_norm": 0.5799348616801011,
"learning_rate": 9.573377856592499e-05,
"loss": 0.8483726978302002,
"step": 294
},
{
"epoch": 0.35520770620108366,
"grad_norm": 0.569975520032633,
"learning_rate": 9.569345359138771e-05,
"loss": 0.6434216499328613,
"step": 295
},
{
"epoch": 0.3564118001204094,
"grad_norm": 0.6057143088843095,
"learning_rate": 9.565294750803575e-05,
"loss": 0.7472547888755798,
"step": 296
},
{
"epoch": 0.3576158940397351,
"grad_norm": 0.5763656433176603,
"learning_rate": 9.561226047641745e-05,
"loss": 0.7766754031181335,
"step": 297
},
{
"epoch": 0.3588199879590608,
"grad_norm": 0.5806884679582975,
"learning_rate": 9.557139265779838e-05,
"loss": 0.7611971497535706,
"step": 298
},
{
"epoch": 0.3600240818783865,
"grad_norm": 0.653182180155357,
"learning_rate": 9.553034421416062e-05,
"loss": 0.8534923195838928,
"step": 299
},
{
"epoch": 0.3612281757977122,
"grad_norm": 0.5930589380494014,
"learning_rate": 9.548911530820226e-05,
"loss": 0.8566128611564636,
"step": 300
},
{
"epoch": 0.3624322697170379,
"grad_norm": 0.594676216957681,
"learning_rate": 9.544770610333655e-05,
"loss": 0.7190215587615967,
"step": 301
},
{
"epoch": 0.36363636363636365,
"grad_norm": 0.5423462580148574,
"learning_rate": 9.540611676369145e-05,
"loss": 0.7619057893753052,
"step": 302
},
{
"epoch": 0.36484045755568933,
"grad_norm": 0.6079659667743106,
"learning_rate": 9.536434745410885e-05,
"loss": 0.9198819398880005,
"step": 303
},
{
"epoch": 0.3660445514750151,
"grad_norm": 0.5481385740366345,
"learning_rate": 9.5322398340144e-05,
"loss": 0.7481440305709839,
"step": 304
},
{
"epoch": 0.36724864539434077,
"grad_norm": 0.5758241389293728,
"learning_rate": 9.528026958806477e-05,
"loss": 0.845780611038208,
"step": 305
},
{
"epoch": 0.36845273931366646,
"grad_norm": 0.6109260118302426,
"learning_rate": 9.523796136485109e-05,
"loss": 0.8439440131187439,
"step": 306
},
{
"epoch": 0.3696568332329922,
"grad_norm": 0.5199941673478126,
"learning_rate": 9.519547383819416e-05,
"loss": 0.8031255006790161,
"step": 307
},
{
"epoch": 0.3708609271523179,
"grad_norm": 0.5886779572113713,
"learning_rate": 9.515280717649594e-05,
"loss": 0.895370364189148,
"step": 308
},
{
"epoch": 0.3720650210716436,
"grad_norm": 0.5241621748424288,
"learning_rate": 9.510996154886839e-05,
"loss": 0.7253029346466064,
"step": 309
},
{
"epoch": 0.3732691149909693,
"grad_norm": 0.5349946958277024,
"learning_rate": 9.506693712513274e-05,
"loss": 0.7455810904502869,
"step": 310
},
{
"epoch": 0.374473208910295,
"grad_norm": 0.5413850636556015,
"learning_rate": 9.502373407581897e-05,
"loss": 0.8433940410614014,
"step": 311
},
{
"epoch": 0.3756773028296207,
"grad_norm": 0.5914855127133216,
"learning_rate": 9.498035257216501e-05,
"loss": 0.8574495911598206,
"step": 312
},
{
"epoch": 0.37688139674894644,
"grad_norm": 0.5786718132723062,
"learning_rate": 9.493679278611616e-05,
"loss": 0.8514014482498169,
"step": 313
},
{
"epoch": 0.37808549066827213,
"grad_norm": 0.677708362567278,
"learning_rate": 9.489305489032425e-05,
"loss": 0.5997920036315918,
"step": 314
},
{
"epoch": 0.3792895845875978,
"grad_norm": 0.6181220022492139,
"learning_rate": 9.484913905814716e-05,
"loss": 0.7272148132324219,
"step": 315
},
{
"epoch": 0.38049367850692356,
"grad_norm": 0.5847159564015513,
"learning_rate": 9.4805045463648e-05,
"loss": 0.7532312870025635,
"step": 316
},
{
"epoch": 0.38169777242624925,
"grad_norm": 0.5884182751349588,
"learning_rate": 9.47607742815944e-05,
"loss": 0.7659265995025635,
"step": 317
},
{
"epoch": 0.38290186634557494,
"grad_norm": 0.5719592491678291,
"learning_rate": 9.471632568745794e-05,
"loss": 0.7821764945983887,
"step": 318
},
{
"epoch": 0.3841059602649007,
"grad_norm": 0.5575576997437829,
"learning_rate": 9.467169985741337e-05,
"loss": 0.800609827041626,
"step": 319
},
{
"epoch": 0.38531005418422637,
"grad_norm": 0.5357479184752278,
"learning_rate": 9.462689696833791e-05,
"loss": 0.713945209980011,
"step": 320
},
{
"epoch": 0.38651414810355206,
"grad_norm": 0.6120104938097841,
"learning_rate": 9.458191719781056e-05,
"loss": 0.8058474063873291,
"step": 321
},
{
"epoch": 0.3877182420228778,
"grad_norm": 0.6693580400917779,
"learning_rate": 9.453676072411142e-05,
"loss": 0.8266781568527222,
"step": 322
},
{
"epoch": 0.3889223359422035,
"grad_norm": 0.5506575412307639,
"learning_rate": 9.449142772622092e-05,
"loss": 0.6587191224098206,
"step": 323
},
{
"epoch": 0.3901264298615292,
"grad_norm": 0.5511777246486614,
"learning_rate": 9.444591838381922e-05,
"loss": 0.7111427187919617,
"step": 324
},
{
"epoch": 0.3913305237808549,
"grad_norm": 0.5746660125461024,
"learning_rate": 9.440023287728537e-05,
"loss": 0.9033454060554504,
"step": 325
},
{
"epoch": 0.3925346177001806,
"grad_norm": 0.569278534242423,
"learning_rate": 9.435437138769672e-05,
"loss": 0.7030330300331116,
"step": 326
},
{
"epoch": 0.3937387116195063,
"grad_norm": 0.6060745891610826,
"learning_rate": 9.430833409682806e-05,
"loss": 0.8240711688995361,
"step": 327
},
{
"epoch": 0.39494280553883204,
"grad_norm": 0.5308427592985828,
"learning_rate": 9.426212118715108e-05,
"loss": 0.735508382320404,
"step": 328
},
{
"epoch": 0.39614689945815773,
"grad_norm": 0.5760677673333868,
"learning_rate": 9.421573284183345e-05,
"loss": 0.8183559775352478,
"step": 329
},
{
"epoch": 0.3973509933774834,
"grad_norm": 0.5071606962993296,
"learning_rate": 9.416916924473825e-05,
"loss": 0.7347877025604248,
"step": 330
},
{
"epoch": 0.39855508729680916,
"grad_norm": 0.5763889971321361,
"learning_rate": 9.412243058042315e-05,
"loss": 0.8280633091926575,
"step": 331
},
{
"epoch": 0.39975918121613485,
"grad_norm": 0.5618383356364371,
"learning_rate": 9.407551703413973e-05,
"loss": 0.7195807695388794,
"step": 332
},
{
"epoch": 0.40096327513546054,
"grad_norm": 0.5895697765110283,
"learning_rate": 9.402842879183272e-05,
"loss": 0.8374162316322327,
"step": 333
},
{
"epoch": 0.4021673690547863,
"grad_norm": 0.5547510447572006,
"learning_rate": 9.398116604013925e-05,
"loss": 0.7269232869148254,
"step": 334
},
{
"epoch": 0.40337146297411197,
"grad_norm": 0.5747636829461152,
"learning_rate": 9.393372896638816e-05,
"loss": 0.7741429805755615,
"step": 335
},
{
"epoch": 0.4045755568934377,
"grad_norm": 0.5582603271143042,
"learning_rate": 9.38861177585992e-05,
"loss": 0.8213120102882385,
"step": 336
},
{
"epoch": 0.4057796508127634,
"grad_norm": 0.5698509481662322,
"learning_rate": 9.383833260548233e-05,
"loss": 0.7737472057342529,
"step": 337
},
{
"epoch": 0.4069837447320891,
"grad_norm": 0.5260228384061386,
"learning_rate": 9.379037369643694e-05,
"loss": 0.8149851560592651,
"step": 338
},
{
"epoch": 0.40818783865141484,
"grad_norm": 0.5765318318304814,
"learning_rate": 9.374224122155111e-05,
"loss": 0.7775697708129883,
"step": 339
},
{
"epoch": 0.4093919325707405,
"grad_norm": 0.49274787307145784,
"learning_rate": 9.369393537160089e-05,
"loss": 0.7359933853149414,
"step": 340
},
{
"epoch": 0.4105960264900662,
"grad_norm": 0.5528181652912268,
"learning_rate": 9.364545633804946e-05,
"loss": 0.6642189025878906,
"step": 341
},
{
"epoch": 0.41180012040939196,
"grad_norm": 0.6207226531988652,
"learning_rate": 9.359680431304647e-05,
"loss": 0.8349613547325134,
"step": 342
},
{
"epoch": 0.41300421432871764,
"grad_norm": 0.6106078516446312,
"learning_rate": 9.354797948942719e-05,
"loss": 0.8384692668914795,
"step": 343
},
{
"epoch": 0.41420830824804333,
"grad_norm": 0.5641755633737662,
"learning_rate": 9.349898206071186e-05,
"loss": 0.8039138317108154,
"step": 344
},
{
"epoch": 0.4154124021673691,
"grad_norm": 0.656469899503437,
"learning_rate": 9.344981222110476e-05,
"loss": 0.8086084723472595,
"step": 345
},
{
"epoch": 0.41661649608669477,
"grad_norm": 0.5864557245393566,
"learning_rate": 9.340047016549358e-05,
"loss": 0.7342469692230225,
"step": 346
},
{
"epoch": 0.41782059000602045,
"grad_norm": 0.6186040249224425,
"learning_rate": 9.335095608944861e-05,
"loss": 0.8007559180259705,
"step": 347
},
{
"epoch": 0.4190246839253462,
"grad_norm": 0.6651504394109262,
"learning_rate": 9.330127018922194e-05,
"loss": 0.7766905426979065,
"step": 348
},
{
"epoch": 0.4202287778446719,
"grad_norm": 0.6545527768938719,
"learning_rate": 9.325141266174666e-05,
"loss": 0.7787045836448669,
"step": 349
},
{
"epoch": 0.4214328717639976,
"grad_norm": 0.5138104578279857,
"learning_rate": 9.320138370463618e-05,
"loss": 0.7259654998779297,
"step": 350
},
{
"epoch": 0.4226369656833233,
"grad_norm": 0.7090214107891576,
"learning_rate": 9.315118351618335e-05,
"loss": 0.8143008947372437,
"step": 351
},
{
"epoch": 0.423841059602649,
"grad_norm": 0.640107824693087,
"learning_rate": 9.310081229535968e-05,
"loss": 0.8052964210510254,
"step": 352
},
{
"epoch": 0.4250451535219747,
"grad_norm": 0.5838874533239824,
"learning_rate": 9.305027024181462e-05,
"loss": 0.7851589322090149,
"step": 353
},
{
"epoch": 0.42624924744130044,
"grad_norm": 0.6875607759241701,
"learning_rate": 9.29995575558747e-05,
"loss": 0.8351097106933594,
"step": 354
},
{
"epoch": 0.4274533413606261,
"grad_norm": 0.6132435281643827,
"learning_rate": 9.294867443854278e-05,
"loss": 0.7207709550857544,
"step": 355
},
{
"epoch": 0.4286574352799518,
"grad_norm": 0.6018423864599579,
"learning_rate": 9.289762109149723e-05,
"loss": 0.7128463983535767,
"step": 356
},
{
"epoch": 0.42986152919927756,
"grad_norm": 0.5966467960285611,
"learning_rate": 9.284639771709112e-05,
"loss": 0.8051083087921143,
"step": 357
},
{
"epoch": 0.43106562311860325,
"grad_norm": 0.5677436973852517,
"learning_rate": 9.279500451835145e-05,
"loss": 0.7338952422142029,
"step": 358
},
{
"epoch": 0.43226971703792894,
"grad_norm": 0.5633327946185779,
"learning_rate": 9.274344169897834e-05,
"loss": 0.7003933787345886,
"step": 359
},
{
"epoch": 0.4334738109572547,
"grad_norm": 0.6099766108022538,
"learning_rate": 9.269170946334418e-05,
"loss": 0.7850499153137207,
"step": 360
},
{
"epoch": 0.43467790487658037,
"grad_norm": 0.6363308217936156,
"learning_rate": 9.263980801649286e-05,
"loss": 0.8188478946685791,
"step": 361
},
{
"epoch": 0.43588199879590606,
"grad_norm": 0.5579250798677722,
"learning_rate": 9.258773756413898e-05,
"loss": 0.731178879737854,
"step": 362
},
{
"epoch": 0.4370860927152318,
"grad_norm": 0.5277326840605376,
"learning_rate": 9.253549831266696e-05,
"loss": 0.7181439399719238,
"step": 363
},
{
"epoch": 0.4382901866345575,
"grad_norm": 0.5991414923556702,
"learning_rate": 9.248309046913032e-05,
"loss": 0.7859128713607788,
"step": 364
},
{
"epoch": 0.4394942805538832,
"grad_norm": 0.5636427682419628,
"learning_rate": 9.243051424125075e-05,
"loss": 0.7138811349868774,
"step": 365
},
{
"epoch": 0.4406983744732089,
"grad_norm": 0.5846552018839858,
"learning_rate": 9.237776983741735e-05,
"loss": 0.8321449756622314,
"step": 366
},
{
"epoch": 0.4419024683925346,
"grad_norm": 0.5961509645070332,
"learning_rate": 9.232485746668584e-05,
"loss": 0.8703262805938721,
"step": 367
},
{
"epoch": 0.44310656231186035,
"grad_norm": 0.6575399410305954,
"learning_rate": 9.227177733877763e-05,
"loss": 0.8056511878967285,
"step": 368
},
{
"epoch": 0.44431065623118604,
"grad_norm": 0.6112157543514062,
"learning_rate": 9.221852966407909e-05,
"loss": 0.81901615858078,
"step": 369
},
{
"epoch": 0.44551475015051173,
"grad_norm": 0.5780705238766984,
"learning_rate": 9.216511465364066e-05,
"loss": 0.7020336389541626,
"step": 370
},
{
"epoch": 0.4467188440698375,
"grad_norm": 0.5513629981168275,
"learning_rate": 9.2111532519176e-05,
"loss": 0.7623274326324463,
"step": 371
},
{
"epoch": 0.44792293798916316,
"grad_norm": 0.5590339331584651,
"learning_rate": 9.205778347306121e-05,
"loss": 0.7354063987731934,
"step": 372
},
{
"epoch": 0.44912703190848885,
"grad_norm": 0.5498974325177253,
"learning_rate": 9.200386772833394e-05,
"loss": 0.6966525316238403,
"step": 373
},
{
"epoch": 0.4503311258278146,
"grad_norm": 0.5404885313697676,
"learning_rate": 9.194978549869256e-05,
"loss": 0.9187078475952148,
"step": 374
},
{
"epoch": 0.4515352197471403,
"grad_norm": 0.574364257307318,
"learning_rate": 9.189553699849531e-05,
"loss": 0.7900419235229492,
"step": 375
},
{
"epoch": 0.45273931366646597,
"grad_norm": 0.5876347961897604,
"learning_rate": 9.184112244275948e-05,
"loss": 0.8052917122840881,
"step": 376
},
{
"epoch": 0.4539434075857917,
"grad_norm": 0.509622094995267,
"learning_rate": 9.178654204716048e-05,
"loss": 0.7819595336914062,
"step": 377
},
{
"epoch": 0.4551475015051174,
"grad_norm": 0.5616258880534972,
"learning_rate": 9.173179602803108e-05,
"loss": 0.8665533065795898,
"step": 378
},
{
"epoch": 0.4563515954244431,
"grad_norm": 0.5674735960986814,
"learning_rate": 9.167688460236049e-05,
"loss": 0.7420907020568848,
"step": 379
},
{
"epoch": 0.45755568934376883,
"grad_norm": 0.5651074504659922,
"learning_rate": 9.162180798779354e-05,
"loss": 0.6834886074066162,
"step": 380
},
{
"epoch": 0.4587597832630945,
"grad_norm": 0.5208113984513062,
"learning_rate": 9.156656640262975e-05,
"loss": 0.7298674583435059,
"step": 381
},
{
"epoch": 0.4599638771824202,
"grad_norm": 0.5568478288456997,
"learning_rate": 9.151116006582259e-05,
"loss": 0.7340203523635864,
"step": 382
},
{
"epoch": 0.46116797110174595,
"grad_norm": 0.6609119540723819,
"learning_rate": 9.145558919697844e-05,
"loss": 0.8512008190155029,
"step": 383
},
{
"epoch": 0.46237206502107164,
"grad_norm": 0.5667512563774485,
"learning_rate": 9.139985401635587e-05,
"loss": 0.7891113758087158,
"step": 384
},
{
"epoch": 0.46357615894039733,
"grad_norm": 0.5542758268779019,
"learning_rate": 9.13439547448647e-05,
"loss": 0.7792179584503174,
"step": 385
},
{
"epoch": 0.4647802528597231,
"grad_norm": 0.5996658004222291,
"learning_rate": 9.128789160406512e-05,
"loss": 0.9395783543586731,
"step": 386
},
{
"epoch": 0.46598434677904876,
"grad_norm": 0.5073928978968797,
"learning_rate": 9.123166481616687e-05,
"loss": 0.6415067911148071,
"step": 387
},
{
"epoch": 0.46718844069837445,
"grad_norm": 0.631792579802824,
"learning_rate": 9.117527460402826e-05,
"loss": 0.8739094734191895,
"step": 388
},
{
"epoch": 0.4683925346177002,
"grad_norm": 0.5554177822601125,
"learning_rate": 9.111872119115537e-05,
"loss": 0.7036592960357666,
"step": 389
},
{
"epoch": 0.4695966285370259,
"grad_norm": 0.5645666409480741,
"learning_rate": 9.106200480170113e-05,
"loss": 0.7435568571090698,
"step": 390
},
{
"epoch": 0.4708007224563516,
"grad_norm": 0.5486147438942627,
"learning_rate": 9.100512566046444e-05,
"loss": 0.6777222752571106,
"step": 391
},
{
"epoch": 0.4720048163756773,
"grad_norm": 0.5789769653823079,
"learning_rate": 9.094808399288927e-05,
"loss": 0.7870496511459351,
"step": 392
},
{
"epoch": 0.473208910295003,
"grad_norm": 0.6306138629757833,
"learning_rate": 9.089088002506379e-05,
"loss": 0.9221470355987549,
"step": 393
},
{
"epoch": 0.4744130042143287,
"grad_norm": 0.63364802954993,
"learning_rate": 9.083351398371944e-05,
"loss": 0.7603942155838013,
"step": 394
},
{
"epoch": 0.47561709813365444,
"grad_norm": 0.7115530757590298,
"learning_rate": 9.077598609623006e-05,
"loss": 0.7661646604537964,
"step": 395
},
{
"epoch": 0.4768211920529801,
"grad_norm": 0.5186352616369874,
"learning_rate": 9.071829659061097e-05,
"loss": 0.7703070640563965,
"step": 396
},
{
"epoch": 0.4780252859723058,
"grad_norm": 0.5523785951320875,
"learning_rate": 9.066044569551807e-05,
"loss": 0.6563143730163574,
"step": 397
},
{
"epoch": 0.47922937989163156,
"grad_norm": 0.6412244640505108,
"learning_rate": 9.060243364024692e-05,
"loss": 0.8129726052284241,
"step": 398
},
{
"epoch": 0.48043347381095725,
"grad_norm": 0.5840224219823051,
"learning_rate": 9.054426065473192e-05,
"loss": 0.7899963855743408,
"step": 399
},
{
"epoch": 0.481637567730283,
"grad_norm": 0.615345356179797,
"learning_rate": 9.048592696954524e-05,
"loss": 0.7919591665267944,
"step": 400
},
{
"epoch": 0.4828416616496087,
"grad_norm": 0.5594464109878114,
"learning_rate": 9.042743281589605e-05,
"loss": 0.7602540254592896,
"step": 401
},
{
"epoch": 0.48404575556893437,
"grad_norm": 0.523704912811271,
"learning_rate": 9.036877842562952e-05,
"loss": 0.8046675324440002,
"step": 402
},
{
"epoch": 0.4852498494882601,
"grad_norm": 0.564482954113039,
"learning_rate": 9.030996403122592e-05,
"loss": 0.8811895847320557,
"step": 403
},
{
"epoch": 0.4864539434075858,
"grad_norm": 0.6386258144233664,
"learning_rate": 9.025098986579975e-05,
"loss": 0.7436861991882324,
"step": 404
},
{
"epoch": 0.4876580373269115,
"grad_norm": 0.5765254802073968,
"learning_rate": 9.019185616309871e-05,
"loss": 0.6859670877456665,
"step": 405
},
{
"epoch": 0.48886213124623723,
"grad_norm": 0.5730166849121991,
"learning_rate": 9.013256315750291e-05,
"loss": 0.7614256143569946,
"step": 406
},
{
"epoch": 0.4900662251655629,
"grad_norm": 0.5357875716257245,
"learning_rate": 9.007311108402379e-05,
"loss": 0.7794307470321655,
"step": 407
},
{
"epoch": 0.4912703190848886,
"grad_norm": 0.5081696602355975,
"learning_rate": 9.001350017830331e-05,
"loss": 0.8063097596168518,
"step": 408
},
{
"epoch": 0.49247441300421435,
"grad_norm": 0.5628512810128503,
"learning_rate": 8.995373067661296e-05,
"loss": 0.8180215358734131,
"step": 409
},
{
"epoch": 0.49367850692354004,
"grad_norm": 0.5787543459084703,
"learning_rate": 8.989380281585287e-05,
"loss": 0.7372187376022339,
"step": 410
},
{
"epoch": 0.4948826008428657,
"grad_norm": 0.5667198954493534,
"learning_rate": 8.983371683355075e-05,
"loss": 0.7635501623153687,
"step": 411
},
{
"epoch": 0.49608669476219147,
"grad_norm": 0.5600547455387459,
"learning_rate": 8.977347296786113e-05,
"loss": 0.7958824634552002,
"step": 412
},
{
"epoch": 0.49729078868151716,
"grad_norm": 0.5625737085034314,
"learning_rate": 8.971307145756426e-05,
"loss": 0.841810405254364,
"step": 413
},
{
"epoch": 0.49849488260084285,
"grad_norm": 0.5614410910946721,
"learning_rate": 8.965251254206524e-05,
"loss": 0.7582192420959473,
"step": 414
},
{
"epoch": 0.4996989765201686,
"grad_norm": 0.5645502655557872,
"learning_rate": 8.959179646139306e-05,
"loss": 0.7968418598175049,
"step": 415
},
{
"epoch": 0.5009030704394943,
"grad_norm": 0.5917884667130379,
"learning_rate": 8.953092345619964e-05,
"loss": 0.7377205491065979,
"step": 416
},
{
"epoch": 0.50210716435882,
"grad_norm": 0.5243909360294917,
"learning_rate": 8.94698937677589e-05,
"loss": 0.7942450642585754,
"step": 417
},
{
"epoch": 0.5033112582781457,
"grad_norm": 0.5980300395240639,
"learning_rate": 8.940870763796574e-05,
"loss": 0.8235912322998047,
"step": 418
},
{
"epoch": 0.5045153521974715,
"grad_norm": 0.5811078832563148,
"learning_rate": 8.934736530933519e-05,
"loss": 0.8105367422103882,
"step": 419
},
{
"epoch": 0.5057194461167971,
"grad_norm": 0.4920507260541056,
"learning_rate": 8.928586702500128e-05,
"loss": 0.7134383916854858,
"step": 420
},
{
"epoch": 0.5069235400361228,
"grad_norm": 0.5367894318907388,
"learning_rate": 8.922421302871629e-05,
"loss": 0.6764764785766602,
"step": 421
},
{
"epoch": 0.5081276339554486,
"grad_norm": 0.5426721151064777,
"learning_rate": 8.916240356484961e-05,
"loss": 0.8036496639251709,
"step": 422
},
{
"epoch": 0.5093317278747742,
"grad_norm": 0.5283738533968033,
"learning_rate": 8.910043887838686e-05,
"loss": 0.7051277160644531,
"step": 423
},
{
"epoch": 0.5105358217941,
"grad_norm": 0.5568815068292188,
"learning_rate": 8.903831921492889e-05,
"loss": 0.7518850564956665,
"step": 424
},
{
"epoch": 0.5117399157134257,
"grad_norm": 0.6019868311120965,
"learning_rate": 8.89760448206908e-05,
"loss": 0.8285863399505615,
"step": 425
},
{
"epoch": 0.5129440096327513,
"grad_norm": 0.5322512751282783,
"learning_rate": 8.8913615942501e-05,
"loss": 0.8142893314361572,
"step": 426
},
{
"epoch": 0.5141481035520771,
"grad_norm": 0.5773593281276044,
"learning_rate": 8.885103282780016e-05,
"loss": 0.7128541469573975,
"step": 427
},
{
"epoch": 0.5153521974714028,
"grad_norm": 0.5265057895614733,
"learning_rate": 8.878829572464034e-05,
"loss": 0.7287251949310303,
"step": 428
},
{
"epoch": 0.5165562913907285,
"grad_norm": 0.6126298896774156,
"learning_rate": 8.872540488168389e-05,
"loss": 0.7118552923202515,
"step": 429
},
{
"epoch": 0.5177603853100542,
"grad_norm": 0.5472240818223483,
"learning_rate": 8.866236054820251e-05,
"loss": 0.7330264449119568,
"step": 430
},
{
"epoch": 0.5189644792293799,
"grad_norm": 0.5423773205241336,
"learning_rate": 8.859916297407636e-05,
"loss": 0.8498375415802002,
"step": 431
},
{
"epoch": 0.5201685731487056,
"grad_norm": 0.511978093072427,
"learning_rate": 8.853581240979285e-05,
"loss": 0.7740522623062134,
"step": 432
},
{
"epoch": 0.5213726670680313,
"grad_norm": 0.6328693375860855,
"learning_rate": 8.847230910644586e-05,
"loss": 0.8013511896133423,
"step": 433
},
{
"epoch": 0.5225767609873571,
"grad_norm": 0.5191934031368447,
"learning_rate": 8.840865331573465e-05,
"loss": 0.8275898098945618,
"step": 434
},
{
"epoch": 0.5237808549066827,
"grad_norm": 0.6078817477329859,
"learning_rate": 8.834484528996287e-05,
"loss": 0.7879576086997986,
"step": 435
},
{
"epoch": 0.5249849488260084,
"grad_norm": 0.5932516211316311,
"learning_rate": 8.828088528203753e-05,
"loss": 0.7822805643081665,
"step": 436
},
{
"epoch": 0.5261890427453342,
"grad_norm": 0.5236270498419456,
"learning_rate": 8.821677354546807e-05,
"loss": 0.8038450479507446,
"step": 437
},
{
"epoch": 0.5273931366646598,
"grad_norm": 0.549932395037056,
"learning_rate": 8.815251033436531e-05,
"loss": 0.734670877456665,
"step": 438
},
{
"epoch": 0.5285972305839856,
"grad_norm": 0.5443522562767869,
"learning_rate": 8.808809590344042e-05,
"loss": 0.7292616367340088,
"step": 439
},
{
"epoch": 0.5298013245033113,
"grad_norm": 0.614247181494645,
"learning_rate": 8.802353050800398e-05,
"loss": 0.8566461801528931,
"step": 440
},
{
"epoch": 0.5310054184226369,
"grad_norm": 0.6195557998692361,
"learning_rate": 8.795881440396491e-05,
"loss": 0.8758805990219116,
"step": 441
},
{
"epoch": 0.5322095123419627,
"grad_norm": 0.6027928486912933,
"learning_rate": 8.789394784782945e-05,
"loss": 0.8482891917228699,
"step": 442
},
{
"epoch": 0.5334136062612884,
"grad_norm": 0.5778637021073555,
"learning_rate": 8.78289310967002e-05,
"loss": 0.8151867389678955,
"step": 443
},
{
"epoch": 0.534617700180614,
"grad_norm": 0.5831694881381801,
"learning_rate": 8.776376440827505e-05,
"loss": 0.7579381465911865,
"step": 444
},
{
"epoch": 0.5358217940999398,
"grad_norm": 0.5875591850025278,
"learning_rate": 8.769844804084619e-05,
"loss": 0.7059892416000366,
"step": 445
},
{
"epoch": 0.5370258880192655,
"grad_norm": 0.5184140330764471,
"learning_rate": 8.763298225329903e-05,
"loss": 0.7210403680801392,
"step": 446
},
{
"epoch": 0.5382299819385912,
"grad_norm": 0.5420847104763425,
"learning_rate": 8.756736730511128e-05,
"loss": 0.7385925650596619,
"step": 447
},
{
"epoch": 0.5394340758579169,
"grad_norm": 0.5935346067614179,
"learning_rate": 8.750160345635183e-05,
"loss": 0.8868709802627563,
"step": 448
},
{
"epoch": 0.5406381697772427,
"grad_norm": 0.5537934394668839,
"learning_rate": 8.74356909676797e-05,
"loss": 0.779586911201477,
"step": 449
},
{
"epoch": 0.5418422636965683,
"grad_norm": 0.5601615221122076,
"learning_rate": 8.736963010034311e-05,
"loss": 0.7346528768539429,
"step": 450
},
{
"epoch": 0.543046357615894,
"grad_norm": 0.5745725564232856,
"learning_rate": 8.73034211161784e-05,
"loss": 0.7848095893859863,
"step": 451
},
{
"epoch": 0.5442504515352198,
"grad_norm": 0.5335933434495655,
"learning_rate": 8.723706427760892e-05,
"loss": 0.7612522840499878,
"step": 452
},
{
"epoch": 0.5454545454545454,
"grad_norm": 0.5227005627709939,
"learning_rate": 8.717055984764411e-05,
"loss": 0.6311644911766052,
"step": 453
},
{
"epoch": 0.5466586393738712,
"grad_norm": 0.569956653853632,
"learning_rate": 8.710390808987833e-05,
"loss": 0.7876021862030029,
"step": 454
},
{
"epoch": 0.5478627332931969,
"grad_norm": 0.5696433799827839,
"learning_rate": 8.703710926848995e-05,
"loss": 0.7248250246047974,
"step": 455
},
{
"epoch": 0.5490668272125225,
"grad_norm": 0.6084782666777745,
"learning_rate": 8.697016364824023e-05,
"loss": 0.816514253616333,
"step": 456
},
{
"epoch": 0.5502709211318483,
"grad_norm": 0.5436677824446347,
"learning_rate": 8.690307149447221e-05,
"loss": 0.7234601974487305,
"step": 457
},
{
"epoch": 0.551475015051174,
"grad_norm": 0.5270803980787487,
"learning_rate": 8.683583307310978e-05,
"loss": 0.6568968892097473,
"step": 458
},
{
"epoch": 0.5526791089704997,
"grad_norm": 0.5707483764552527,
"learning_rate": 8.676844865065659e-05,
"loss": 0.777597188949585,
"step": 459
},
{
"epoch": 0.5538832028898254,
"grad_norm": 0.640463873721775,
"learning_rate": 8.67009184941949e-05,
"loss": 0.9555122256278992,
"step": 460
},
{
"epoch": 0.5550872968091511,
"grad_norm": 0.5567654676409305,
"learning_rate": 8.663324287138469e-05,
"loss": 0.8270890712738037,
"step": 461
},
{
"epoch": 0.5562913907284768,
"grad_norm": 0.5072057355413937,
"learning_rate": 8.656542205046243e-05,
"loss": 0.6548333764076233,
"step": 462
},
{
"epoch": 0.5574954846478025,
"grad_norm": 0.5192512231836927,
"learning_rate": 8.64974563002401e-05,
"loss": 0.7468447089195251,
"step": 463
},
{
"epoch": 0.5586995785671283,
"grad_norm": 0.6679093927996793,
"learning_rate": 8.642934589010414e-05,
"loss": 0.8109292387962341,
"step": 464
},
{
"epoch": 0.5599036724864539,
"grad_norm": 0.566273727377757,
"learning_rate": 8.636109109001438e-05,
"loss": 0.7278716564178467,
"step": 465
},
{
"epoch": 0.5611077664057796,
"grad_norm": 0.5609594487670674,
"learning_rate": 8.629269217050289e-05,
"loss": 0.8199411630630493,
"step": 466
},
{
"epoch": 0.5623118603251054,
"grad_norm": 0.5367001348544211,
"learning_rate": 8.6224149402673e-05,
"loss": 0.7463403344154358,
"step": 467
},
{
"epoch": 0.563515954244431,
"grad_norm": 0.546359613860392,
"learning_rate": 8.61554630581982e-05,
"loss": 0.7412172555923462,
"step": 468
},
{
"epoch": 0.5647200481637568,
"grad_norm": 0.5641710231968654,
"learning_rate": 8.608663340932104e-05,
"loss": 0.758033275604248,
"step": 469
},
{
"epoch": 0.5659241420830825,
"grad_norm": 0.5512535825337246,
"learning_rate": 8.601766072885204e-05,
"loss": 0.8045417070388794,
"step": 470
},
{
"epoch": 0.5671282360024081,
"grad_norm": 0.5665039364926218,
"learning_rate": 8.594854529016872e-05,
"loss": 0.7329437732696533,
"step": 471
},
{
"epoch": 0.5683323299217339,
"grad_norm": 0.5888725079503595,
"learning_rate": 8.587928736721432e-05,
"loss": 0.669249951839447,
"step": 472
},
{
"epoch": 0.5695364238410596,
"grad_norm": 0.6183004106038696,
"learning_rate": 8.580988723449688e-05,
"loss": 0.8152425289154053,
"step": 473
},
{
"epoch": 0.5707405177603853,
"grad_norm": 0.5391269856377908,
"learning_rate": 8.574034516708814e-05,
"loss": 0.7069454789161682,
"step": 474
},
{
"epoch": 0.571944611679711,
"grad_norm": 0.543803781167708,
"learning_rate": 8.567066144062232e-05,
"loss": 0.819575309753418,
"step": 475
},
{
"epoch": 0.5731487055990367,
"grad_norm": 0.5310954607891327,
"learning_rate": 8.560083633129519e-05,
"loss": 0.7727447748184204,
"step": 476
},
{
"epoch": 0.5743527995183624,
"grad_norm": 0.5336411225614897,
"learning_rate": 8.553087011586284e-05,
"loss": 0.7410896420478821,
"step": 477
},
{
"epoch": 0.5755568934376881,
"grad_norm": 0.5160334921971708,
"learning_rate": 8.546076307164068e-05,
"loss": 0.6553800106048584,
"step": 478
},
{
"epoch": 0.5767609873570139,
"grad_norm": 0.5302079661582317,
"learning_rate": 8.53905154765023e-05,
"loss": 0.7029821872711182,
"step": 479
},
{
"epoch": 0.5779650812763396,
"grad_norm": 0.5600636568683036,
"learning_rate": 8.532012760887837e-05,
"loss": 0.7383387088775635,
"step": 480
},
{
"epoch": 0.5791691751956652,
"grad_norm": 0.5998929893076258,
"learning_rate": 8.524959974775551e-05,
"loss": 0.6950462460517883,
"step": 481
},
{
"epoch": 0.580373269114991,
"grad_norm": 0.5365329118532681,
"learning_rate": 8.517893217267525e-05,
"loss": 0.778092622756958,
"step": 482
},
{
"epoch": 0.5815773630343167,
"grad_norm": 0.5368638703097667,
"learning_rate": 8.510812516373288e-05,
"loss": 0.7105111479759216,
"step": 483
},
{
"epoch": 0.5827814569536424,
"grad_norm": 0.5907429035845754,
"learning_rate": 8.503717900157632e-05,
"loss": 0.7242652177810669,
"step": 484
},
{
"epoch": 0.5839855508729681,
"grad_norm": 0.5536338375378821,
"learning_rate": 8.496609396740506e-05,
"loss": 0.7318590879440308,
"step": 485
},
{
"epoch": 0.5851896447922939,
"grad_norm": 0.5372423258283444,
"learning_rate": 8.489487034296902e-05,
"loss": 0.693002462387085,
"step": 486
},
{
"epoch": 0.5863937387116195,
"grad_norm": 0.5789242229583803,
"learning_rate": 8.482350841056737e-05,
"loss": 0.8206942081451416,
"step": 487
},
{
"epoch": 0.5875978326309452,
"grad_norm": 0.5233343970495764,
"learning_rate": 8.475200845304758e-05,
"loss": 0.7247848510742188,
"step": 488
},
{
"epoch": 0.588801926550271,
"grad_norm": 0.5660226782273997,
"learning_rate": 8.468037075380408e-05,
"loss": 0.7088282108306885,
"step": 489
},
{
"epoch": 0.5900060204695966,
"grad_norm": 0.6290332733728029,
"learning_rate": 8.460859559677734e-05,
"loss": 0.7779991626739502,
"step": 490
},
{
"epoch": 0.5912101143889223,
"grad_norm": 0.5914880251961212,
"learning_rate": 8.453668326645259e-05,
"loss": 0.6892540454864502,
"step": 491
},
{
"epoch": 0.5924142083082481,
"grad_norm": 0.5921474437264784,
"learning_rate": 8.446463404785875e-05,
"loss": 0.7645236253738403,
"step": 492
},
{
"epoch": 0.5936183022275737,
"grad_norm": 0.527891720939569,
"learning_rate": 8.43924482265674e-05,
"loss": 0.699411153793335,
"step": 493
},
{
"epoch": 0.5948223961468995,
"grad_norm": 0.5808472585843106,
"learning_rate": 8.432012608869141e-05,
"loss": 0.6047768592834473,
"step": 494
},
{
"epoch": 0.5960264900662252,
"grad_norm": 0.7262338174416691,
"learning_rate": 8.424766792088408e-05,
"loss": 0.7811301946640015,
"step": 495
},
{
"epoch": 0.5972305839855508,
"grad_norm": 0.6179024400861065,
"learning_rate": 8.417507401033779e-05,
"loss": 0.7106954455375671,
"step": 496
},
{
"epoch": 0.5984346779048766,
"grad_norm": 0.5815438691669637,
"learning_rate": 8.410234464478297e-05,
"loss": 0.7327958345413208,
"step": 497
},
{
"epoch": 0.5996387718242023,
"grad_norm": 0.5730949644157526,
"learning_rate": 8.402948011248692e-05,
"loss": 0.7256393432617188,
"step": 498
},
{
"epoch": 0.600842865743528,
"grad_norm": 0.521591136634977,
"learning_rate": 8.395648070225272e-05,
"loss": 0.7140817046165466,
"step": 499
},
{
"epoch": 0.6020469596628537,
"grad_norm": 0.5589447096353005,
"learning_rate": 8.388334670341805e-05,
"loss": 0.8448606729507446,
"step": 500
},
{
"epoch": 0.6032510535821795,
"grad_norm": 0.6149413795804055,
"learning_rate": 8.381007840585395e-05,
"loss": 0.8227097988128662,
"step": 501
},
{
"epoch": 0.6044551475015051,
"grad_norm": 0.5875235914864911,
"learning_rate": 8.373667609996387e-05,
"loss": 0.7119932174682617,
"step": 502
},
{
"epoch": 0.6056592414208308,
"grad_norm": 0.577257899771382,
"learning_rate": 8.366314007668235e-05,
"loss": 0.8036626577377319,
"step": 503
},
{
"epoch": 0.6068633353401566,
"grad_norm": 0.6029845794394658,
"learning_rate": 8.358947062747397e-05,
"loss": 0.7943709492683411,
"step": 504
},
{
"epoch": 0.6080674292594822,
"grad_norm": 0.5679689315213442,
"learning_rate": 8.351566804433207e-05,
"loss": 0.7255787253379822,
"step": 505
},
{
"epoch": 0.609271523178808,
"grad_norm": 0.5861915732495735,
"learning_rate": 8.344173261977777e-05,
"loss": 0.7799243927001953,
"step": 506
},
{
"epoch": 0.6104756170981337,
"grad_norm": 0.5291794657388438,
"learning_rate": 8.336766464685869e-05,
"loss": 0.6609664559364319,
"step": 507
},
{
"epoch": 0.6116797110174593,
"grad_norm": 0.5419697178161531,
"learning_rate": 8.329346441914774e-05,
"loss": 0.7047343850135803,
"step": 508
},
{
"epoch": 0.6128838049367851,
"grad_norm": 0.5805226842218582,
"learning_rate": 8.321913223074212e-05,
"loss": 0.8172140121459961,
"step": 509
},
{
"epoch": 0.6140878988561108,
"grad_norm": 0.5298500503507028,
"learning_rate": 8.314466837626205e-05,
"loss": 0.7412742376327515,
"step": 510
},
{
"epoch": 0.6152919927754364,
"grad_norm": 0.5840010321657718,
"learning_rate": 8.307007315084958e-05,
"loss": 0.8661916851997375,
"step": 511
},
{
"epoch": 0.6164960866947622,
"grad_norm": 0.5728057698967226,
"learning_rate": 8.299534685016747e-05,
"loss": 0.8871906399726868,
"step": 512
},
{
"epoch": 0.6177001806140879,
"grad_norm": 0.5253990819928123,
"learning_rate": 8.292048977039801e-05,
"loss": 0.7414618730545044,
"step": 513
},
{
"epoch": 0.6189042745334136,
"grad_norm": 0.5776425296105538,
"learning_rate": 8.284550220824187e-05,
"loss": 0.8262412548065186,
"step": 514
},
{
"epoch": 0.6201083684527393,
"grad_norm": 0.5631058348593229,
"learning_rate": 8.277038446091683e-05,
"loss": 0.7710633277893066,
"step": 515
},
{
"epoch": 0.621312462372065,
"grad_norm": 0.575180578671768,
"learning_rate": 8.269513682615672e-05,
"loss": 0.7422937750816345,
"step": 516
},
{
"epoch": 0.6225165562913907,
"grad_norm": 0.5628819507888372,
"learning_rate": 8.261975960221017e-05,
"loss": 0.7483841776847839,
"step": 517
},
{
"epoch": 0.6237206502107164,
"grad_norm": 0.5492417710470213,
"learning_rate": 8.254425308783944e-05,
"loss": 0.735903799533844,
"step": 518
},
{
"epoch": 0.6249247441300422,
"grad_norm": 0.6072331817997367,
"learning_rate": 8.246861758231925e-05,
"loss": 0.8859599828720093,
"step": 519
},
{
"epoch": 0.6261288380493678,
"grad_norm": 0.5334883775025814,
"learning_rate": 8.239285338543558e-05,
"loss": 0.8330262303352356,
"step": 520
},
{
"epoch": 0.6273329319686936,
"grad_norm": 0.5255759564099236,
"learning_rate": 8.23169607974845e-05,
"loss": 0.7606721520423889,
"step": 521
},
{
"epoch": 0.6285370258880193,
"grad_norm": 0.5114246112951885,
"learning_rate": 8.224094011927091e-05,
"loss": 0.6729496717453003,
"step": 522
},
{
"epoch": 0.6297411198073449,
"grad_norm": 0.6173305080246639,
"learning_rate": 8.216479165210748e-05,
"loss": 0.7409894466400146,
"step": 523
},
{
"epoch": 0.6309452137266707,
"grad_norm": 0.5617836428921698,
"learning_rate": 8.208851569781335e-05,
"loss": 0.7701466083526611,
"step": 524
},
{
"epoch": 0.6321493076459964,
"grad_norm": 0.556774285143137,
"learning_rate": 8.201211255871293e-05,
"loss": 0.7755535840988159,
"step": 525
},
{
"epoch": 0.633353401565322,
"grad_norm": 0.5473181136107016,
"learning_rate": 8.193558253763478e-05,
"loss": 0.771713137626648,
"step": 526
},
{
"epoch": 0.6345574954846478,
"grad_norm": 0.5892189910907214,
"learning_rate": 8.185892593791034e-05,
"loss": 0.8112677335739136,
"step": 527
},
{
"epoch": 0.6357615894039735,
"grad_norm": 0.49813360212693836,
"learning_rate": 8.178214306337278e-05,
"loss": 0.6962313652038574,
"step": 528
},
{
"epoch": 0.6369656833232992,
"grad_norm": 0.5163153013576592,
"learning_rate": 8.170523421835572e-05,
"loss": 0.6276844143867493,
"step": 529
},
{
"epoch": 0.6381697772426249,
"grad_norm": 0.5764131088706974,
"learning_rate": 8.162819970769211e-05,
"loss": 0.8286241888999939,
"step": 530
},
{
"epoch": 0.6393738711619507,
"grad_norm": 0.5345116258038959,
"learning_rate": 8.155103983671297e-05,
"loss": 0.6163603067398071,
"step": 531
},
{
"epoch": 0.6405779650812763,
"grad_norm": 0.5693561222753533,
"learning_rate": 8.14737549112462e-05,
"loss": 0.7609665989875793,
"step": 532
},
{
"epoch": 0.641782059000602,
"grad_norm": 0.5363204766451696,
"learning_rate": 8.139634523761537e-05,
"loss": 0.7703006863594055,
"step": 533
},
{
"epoch": 0.6429861529199278,
"grad_norm": 0.5018955837394068,
"learning_rate": 8.131881112263845e-05,
"loss": 0.7416508793830872,
"step": 534
},
{
"epoch": 0.6441902468392534,
"grad_norm": 0.5609638717660475,
"learning_rate": 8.12411528736267e-05,
"loss": 0.8192243576049805,
"step": 535
},
{
"epoch": 0.6453943407585792,
"grad_norm": 0.6780705920762121,
"learning_rate": 8.116337079838337e-05,
"loss": 0.7729198932647705,
"step": 536
},
{
"epoch": 0.6465984346779049,
"grad_norm": 0.5440984121290051,
"learning_rate": 8.10854652052025e-05,
"loss": 0.6821568012237549,
"step": 537
},
{
"epoch": 0.6478025285972305,
"grad_norm": 0.5717350810710682,
"learning_rate": 8.100743640286768e-05,
"loss": 0.7128407955169678,
"step": 538
},
{
"epoch": 0.6490066225165563,
"grad_norm": 0.5284572857325969,
"learning_rate": 8.092928470065091e-05,
"loss": 0.7074618339538574,
"step": 539
},
{
"epoch": 0.650210716435882,
"grad_norm": 0.5074640985182344,
"learning_rate": 8.085101040831122e-05,
"loss": 0.646845817565918,
"step": 540
},
{
"epoch": 0.6514148103552077,
"grad_norm": 0.5543246780629942,
"learning_rate": 8.077261383609363e-05,
"loss": 0.773701548576355,
"step": 541
},
{
"epoch": 0.6526189042745334,
"grad_norm": 0.6160660268489386,
"learning_rate": 8.069409529472774e-05,
"loss": 0.6822642683982849,
"step": 542
},
{
"epoch": 0.6538229981938591,
"grad_norm": 0.53725601622277,
"learning_rate": 8.061545509542663e-05,
"loss": 0.706203818321228,
"step": 543
},
{
"epoch": 0.6550270921131849,
"grad_norm": 0.5298470910298363,
"learning_rate": 8.05366935498856e-05,
"loss": 0.7028369903564453,
"step": 544
},
{
"epoch": 0.6562311860325105,
"grad_norm": 0.5289295675125671,
"learning_rate": 8.045781097028083e-05,
"loss": 0.8363953232765198,
"step": 545
},
{
"epoch": 0.6574352799518363,
"grad_norm": 0.5484514846314679,
"learning_rate": 8.037880766926833e-05,
"loss": 0.8546597361564636,
"step": 546
},
{
"epoch": 0.658639373871162,
"grad_norm": 0.5743868068932564,
"learning_rate": 8.02996839599825e-05,
"loss": 0.7040009498596191,
"step": 547
},
{
"epoch": 0.6598434677904876,
"grad_norm": 0.5717016239116167,
"learning_rate": 8.022044015603505e-05,
"loss": 0.78245609998703,
"step": 548
},
{
"epoch": 0.6610475617098134,
"grad_norm": 0.5863267636460782,
"learning_rate": 8.014107657151369e-05,
"loss": 0.8965244293212891,
"step": 549
},
{
"epoch": 0.6622516556291391,
"grad_norm": 0.5058811889612245,
"learning_rate": 8.006159352098082e-05,
"loss": 0.7318023443222046,
"step": 550
},
{
"epoch": 0.6634557495484648,
"grad_norm": 0.5481044361172038,
"learning_rate": 7.998199131947247e-05,
"loss": 0.8633524775505066,
"step": 551
},
{
"epoch": 0.6646598434677905,
"grad_norm": 0.5693440196646484,
"learning_rate": 7.990227028249678e-05,
"loss": 0.73511803150177,
"step": 552
},
{
"epoch": 0.6658639373871162,
"grad_norm": 0.5738698013859286,
"learning_rate": 7.982243072603306e-05,
"loss": 0.7095030546188354,
"step": 553
},
{
"epoch": 0.6670680313064419,
"grad_norm": 0.5798536139502265,
"learning_rate": 7.974247296653028e-05,
"loss": 0.6944142580032349,
"step": 554
},
{
"epoch": 0.6682721252257676,
"grad_norm": 0.5445345284588756,
"learning_rate": 7.966239732090592e-05,
"loss": 0.7495391368865967,
"step": 555
},
{
"epoch": 0.6694762191450934,
"grad_norm": 0.5038971804023679,
"learning_rate": 7.958220410654475e-05,
"loss": 0.6984772086143494,
"step": 556
},
{
"epoch": 0.670680313064419,
"grad_norm": 0.5600022210005443,
"learning_rate": 7.95018936412975e-05,
"loss": 0.7748847007751465,
"step": 557
},
{
"epoch": 0.6718844069837447,
"grad_norm": 0.5147765169482994,
"learning_rate": 7.942146624347964e-05,
"loss": 0.6932211518287659,
"step": 558
},
{
"epoch": 0.6730885009030705,
"grad_norm": 0.5425861579492259,
"learning_rate": 7.93409222318701e-05,
"loss": 0.7832596302032471,
"step": 559
},
{
"epoch": 0.6742925948223961,
"grad_norm": 0.5209359479363717,
"learning_rate": 7.926026192571007e-05,
"loss": 0.7210221290588379,
"step": 560
},
{
"epoch": 0.6754966887417219,
"grad_norm": 0.5337914370112163,
"learning_rate": 7.917948564470157e-05,
"loss": 0.78125,
"step": 561
},
{
"epoch": 0.6767007826610476,
"grad_norm": 0.5630510669001023,
"learning_rate": 7.909859370900642e-05,
"loss": 0.656562089920044,
"step": 562
},
{
"epoch": 0.6779048765803732,
"grad_norm": 0.6145722959937302,
"learning_rate": 7.901758643924475e-05,
"loss": 0.7918832302093506,
"step": 563
},
{
"epoch": 0.679108970499699,
"grad_norm": 0.5595323369440784,
"learning_rate": 7.893646415649384e-05,
"loss": 0.6872172355651855,
"step": 564
},
{
"epoch": 0.6803130644190247,
"grad_norm": 0.559585873258208,
"learning_rate": 7.88552271822869e-05,
"loss": 0.7413808107376099,
"step": 565
},
{
"epoch": 0.6815171583383504,
"grad_norm": 0.5896344593158469,
"learning_rate": 7.877387583861165e-05,
"loss": 0.7586731910705566,
"step": 566
},
{
"epoch": 0.6827212522576761,
"grad_norm": 0.7785143302301188,
"learning_rate": 7.869241044790915e-05,
"loss": 0.8225888609886169,
"step": 567
},
{
"epoch": 0.6839253461770018,
"grad_norm": 0.5216852463423214,
"learning_rate": 7.861083133307247e-05,
"loss": 0.7068048715591431,
"step": 568
},
{
"epoch": 0.6851294400963275,
"grad_norm": 0.546179315042222,
"learning_rate": 7.852913881744547e-05,
"loss": 0.7423121333122253,
"step": 569
},
{
"epoch": 0.6863335340156532,
"grad_norm": 0.603434739035308,
"learning_rate": 7.844733322482145e-05,
"loss": 0.7143921256065369,
"step": 570
},
{
"epoch": 0.687537627934979,
"grad_norm": 0.5559389093356377,
"learning_rate": 7.836541487944193e-05,
"loss": 0.7010910511016846,
"step": 571
},
{
"epoch": 0.6887417218543046,
"grad_norm": 0.5487579113202334,
"learning_rate": 7.82833841059953e-05,
"loss": 0.7683477401733398,
"step": 572
},
{
"epoch": 0.6899458157736303,
"grad_norm": 0.5395639066708008,
"learning_rate": 7.820124122961557e-05,
"loss": 0.663609504699707,
"step": 573
},
{
"epoch": 0.6911499096929561,
"grad_norm": 0.5384530868694132,
"learning_rate": 7.811898657588109e-05,
"loss": 0.6645287275314331,
"step": 574
},
{
"epoch": 0.6923540036122817,
"grad_norm": 0.5220245375313282,
"learning_rate": 7.803662047081323e-05,
"loss": 0.741969883441925,
"step": 575
},
{
"epoch": 0.6935580975316075,
"grad_norm": 0.5126476994514748,
"learning_rate": 7.795414324087515e-05,
"loss": 0.7474612593650818,
"step": 576
},
{
"epoch": 0.6947621914509332,
"grad_norm": 0.5328593914899424,
"learning_rate": 7.78715552129704e-05,
"loss": 0.8475122451782227,
"step": 577
},
{
"epoch": 0.6959662853702588,
"grad_norm": 0.5116580634330516,
"learning_rate": 7.778885671444169e-05,
"loss": 0.5925725698471069,
"step": 578
},
{
"epoch": 0.6971703792895846,
"grad_norm": 0.5209404691403938,
"learning_rate": 7.770604807306966e-05,
"loss": 0.7660278081893921,
"step": 579
},
{
"epoch": 0.6983744732089103,
"grad_norm": 0.6419678559064697,
"learning_rate": 7.762312961707141e-05,
"loss": 0.8105254769325256,
"step": 580
},
{
"epoch": 0.699578567128236,
"grad_norm": 0.5291551771982791,
"learning_rate": 7.754010167509935e-05,
"loss": 0.7684112191200256,
"step": 581
},
{
"epoch": 0.7007826610475617,
"grad_norm": 0.5344021885735301,
"learning_rate": 7.745696457623986e-05,
"loss": 0.7174640893936157,
"step": 582
},
{
"epoch": 0.7019867549668874,
"grad_norm": 0.5768957782540212,
"learning_rate": 7.73737186500119e-05,
"loss": 0.8031843304634094,
"step": 583
},
{
"epoch": 0.7031908488862131,
"grad_norm": 0.5426844722988857,
"learning_rate": 7.729036422636589e-05,
"loss": 0.7134600281715393,
"step": 584
},
{
"epoch": 0.7043949428055388,
"grad_norm": 0.5030850249501789,
"learning_rate": 7.720690163568214e-05,
"loss": 0.6457868218421936,
"step": 585
},
{
"epoch": 0.7055990367248646,
"grad_norm": 0.6191772999257755,
"learning_rate": 7.712333120876983e-05,
"loss": 0.7789636850357056,
"step": 586
},
{
"epoch": 0.7068031306441902,
"grad_norm": 0.5321381054419203,
"learning_rate": 7.703965327686544e-05,
"loss": 0.6850986480712891,
"step": 587
},
{
"epoch": 0.708007224563516,
"grad_norm": 0.525345105210121,
"learning_rate": 7.695586817163163e-05,
"loss": 0.7084580659866333,
"step": 588
},
{
"epoch": 0.7092113184828417,
"grad_norm": 0.5395663278470533,
"learning_rate": 7.68719762251558e-05,
"loss": 0.6684824824333191,
"step": 589
},
{
"epoch": 0.7104154124021673,
"grad_norm": 0.5825678154513388,
"learning_rate": 7.678797776994886e-05,
"loss": 0.671932578086853,
"step": 590
},
{
"epoch": 0.7116195063214931,
"grad_norm": 0.5681195138990598,
"learning_rate": 7.670387313894384e-05,
"loss": 0.7614145278930664,
"step": 591
},
{
"epoch": 0.7128236002408188,
"grad_norm": 0.5747985438529684,
"learning_rate": 7.661966266549463e-05,
"loss": 0.6186734437942505,
"step": 592
},
{
"epoch": 0.7140276941601444,
"grad_norm": 0.615632973884984,
"learning_rate": 7.653534668337463e-05,
"loss": 0.7583650350570679,
"step": 593
},
{
"epoch": 0.7152317880794702,
"grad_norm": 0.5845374532491585,
"learning_rate": 7.645092552677539e-05,
"loss": 0.7116202116012573,
"step": 594
},
{
"epoch": 0.7164358819987959,
"grad_norm": 0.5194310981235325,
"learning_rate": 7.636639953030541e-05,
"loss": 0.724486231803894,
"step": 595
},
{
"epoch": 0.7176399759181216,
"grad_norm": 0.5271839381183305,
"learning_rate": 7.628176902898863e-05,
"loss": 0.7952111959457397,
"step": 596
},
{
"epoch": 0.7188440698374473,
"grad_norm": 0.5466556917732034,
"learning_rate": 7.619703435826328e-05,
"loss": 0.8017375469207764,
"step": 597
},
{
"epoch": 0.720048163756773,
"grad_norm": 0.5805731516392539,
"learning_rate": 7.61121958539804e-05,
"loss": 0.7721456289291382,
"step": 598
},
{
"epoch": 0.7212522576760987,
"grad_norm": 0.5750677094113634,
"learning_rate": 7.602725385240268e-05,
"loss": 0.8046045303344727,
"step": 599
},
{
"epoch": 0.7224563515954244,
"grad_norm": 0.5050120132500692,
"learning_rate": 7.594220869020293e-05,
"loss": 0.718019962310791,
"step": 600
},
{
"epoch": 0.7236604455147502,
"grad_norm": 0.5775475591845226,
"learning_rate": 7.585706070446288e-05,
"loss": 0.681684672832489,
"step": 601
},
{
"epoch": 0.7248645394340758,
"grad_norm": 0.5650253516457611,
"learning_rate": 7.577181023267185e-05,
"loss": 0.8424392342567444,
"step": 602
},
{
"epoch": 0.7260686333534015,
"grad_norm": 0.5597171944868315,
"learning_rate": 7.568645761272527e-05,
"loss": 0.7912712693214417,
"step": 603
},
{
"epoch": 0.7272727272727273,
"grad_norm": 0.6549911820787678,
"learning_rate": 7.560100318292355e-05,
"loss": 0.7041141390800476,
"step": 604
},
{
"epoch": 0.7284768211920529,
"grad_norm": 0.5358008356978227,
"learning_rate": 7.551544728197057e-05,
"loss": 0.7670320272445679,
"step": 605
},
{
"epoch": 0.7296809151113787,
"grad_norm": 0.4901100224928788,
"learning_rate": 7.542979024897239e-05,
"loss": 0.7009145021438599,
"step": 606
},
{
"epoch": 0.7308850090307044,
"grad_norm": 0.5905232471499013,
"learning_rate": 7.534403242343595e-05,
"loss": 0.7351090312004089,
"step": 607
},
{
"epoch": 0.7320891029500302,
"grad_norm": 0.5500568361307169,
"learning_rate": 7.525817414526764e-05,
"loss": 0.7852768301963806,
"step": 608
},
{
"epoch": 0.7332931968693558,
"grad_norm": 0.5278365030547066,
"learning_rate": 7.517221575477209e-05,
"loss": 0.7615916728973389,
"step": 609
},
{
"epoch": 0.7344972907886815,
"grad_norm": 0.5586068544844466,
"learning_rate": 7.508615759265059e-05,
"loss": 0.7317056655883789,
"step": 610
},
{
"epoch": 0.7357013847080073,
"grad_norm": 0.5567556937034017,
"learning_rate": 7.500000000000001e-05,
"loss": 0.6546356678009033,
"step": 611
},
{
"epoch": 0.7369054786273329,
"grad_norm": 0.5190616958413415,
"learning_rate": 7.491374331831125e-05,
"loss": 0.6745877861976624,
"step": 612
},
{
"epoch": 0.7381095725466587,
"grad_norm": 0.5155477634721394,
"learning_rate": 7.482738788946799e-05,
"loss": 0.6520372629165649,
"step": 613
},
{
"epoch": 0.7393136664659844,
"grad_norm": 0.5335041167853599,
"learning_rate": 7.474093405574527e-05,
"loss": 0.7490137219429016,
"step": 614
},
{
"epoch": 0.74051776038531,
"grad_norm": 0.7109509898226887,
"learning_rate": 7.465438215980819e-05,
"loss": 0.7756985425949097,
"step": 615
},
{
"epoch": 0.7417218543046358,
"grad_norm": 0.5955093151487647,
"learning_rate": 7.456773254471053e-05,
"loss": 0.7039395570755005,
"step": 616
},
{
"epoch": 0.7429259482239615,
"grad_norm": 1.7500643108130063,
"learning_rate": 7.448098555389333e-05,
"loss": 0.7679908871650696,
"step": 617
},
{
"epoch": 0.7441300421432872,
"grad_norm": 0.5689917526414457,
"learning_rate": 7.439414153118364e-05,
"loss": 0.7280757427215576,
"step": 618
},
{
"epoch": 0.7453341360626129,
"grad_norm": 0.5541120701637754,
"learning_rate": 7.43072008207931e-05,
"loss": 0.7403719425201416,
"step": 619
},
{
"epoch": 0.7465382299819386,
"grad_norm": 0.5572484684641944,
"learning_rate": 7.422016376731658e-05,
"loss": 0.6732151508331299,
"step": 620
},
{
"epoch": 0.7477423239012643,
"grad_norm": 0.6112781170222843,
"learning_rate": 7.413303071573077e-05,
"loss": 0.7835999727249146,
"step": 621
},
{
"epoch": 0.74894641782059,
"grad_norm": 0.5389600618950466,
"learning_rate": 7.404580201139286e-05,
"loss": 0.7648690938949585,
"step": 622
},
{
"epoch": 0.7501505117399158,
"grad_norm": 0.5604655172081615,
"learning_rate": 7.395847800003925e-05,
"loss": 0.7572316527366638,
"step": 623
},
{
"epoch": 0.7513546056592414,
"grad_norm": 0.6075035154379582,
"learning_rate": 7.387105902778397e-05,
"loss": 0.7459697127342224,
"step": 624
},
{
"epoch": 0.7525586995785671,
"grad_norm": 0.5278952094466931,
"learning_rate": 7.378354544111755e-05,
"loss": 0.6718525290489197,
"step": 625
},
{
"epoch": 0.7537627934978929,
"grad_norm": 0.5584227775244037,
"learning_rate": 7.36959375869054e-05,
"loss": 0.8716173768043518,
"step": 626
},
{
"epoch": 0.7549668874172185,
"grad_norm": 0.5530155578773712,
"learning_rate": 7.360823581238672e-05,
"loss": 0.7087276577949524,
"step": 627
},
{
"epoch": 0.7561709813365443,
"grad_norm": 0.5059165745763609,
"learning_rate": 7.352044046517285e-05,
"loss": 0.6190001964569092,
"step": 628
},
{
"epoch": 0.75737507525587,
"grad_norm": 0.5098108817484567,
"learning_rate": 7.343255189324605e-05,
"loss": 0.7209186553955078,
"step": 629
},
{
"epoch": 0.7585791691751956,
"grad_norm": 0.5806534819653945,
"learning_rate": 7.334457044495811e-05,
"loss": 0.7904737591743469,
"step": 630
},
{
"epoch": 0.7597832630945214,
"grad_norm": 0.5678941011790182,
"learning_rate": 7.325649646902887e-05,
"loss": 0.7746214270591736,
"step": 631
},
{
"epoch": 0.7609873570138471,
"grad_norm": 0.53499301398341,
"learning_rate": 7.316833031454498e-05,
"loss": 0.81767737865448,
"step": 632
},
{
"epoch": 0.7621914509331728,
"grad_norm": 0.6293538353705853,
"learning_rate": 7.30800723309584e-05,
"loss": 0.754915177822113,
"step": 633
},
{
"epoch": 0.7633955448524985,
"grad_norm": 0.5024549487657861,
"learning_rate": 7.299172286808511e-05,
"loss": 0.6980667114257812,
"step": 634
},
{
"epoch": 0.7645996387718242,
"grad_norm": 0.5808912591228869,
"learning_rate": 7.290328227610362e-05,
"loss": 0.7346571087837219,
"step": 635
},
{
"epoch": 0.7658037326911499,
"grad_norm": 0.5040533014385249,
"learning_rate": 7.281475090555365e-05,
"loss": 0.7573859095573425,
"step": 636
},
{
"epoch": 0.7670078266104756,
"grad_norm": 0.6000308844739162,
"learning_rate": 7.272612910733475e-05,
"loss": 0.734082818031311,
"step": 637
},
{
"epoch": 0.7682119205298014,
"grad_norm": 0.5704625126191725,
"learning_rate": 7.263741723270486e-05,
"loss": 0.7449295520782471,
"step": 638
},
{
"epoch": 0.769416014449127,
"grad_norm": 0.5466539737703434,
"learning_rate": 7.254861563327896e-05,
"loss": 0.6105548143386841,
"step": 639
},
{
"epoch": 0.7706201083684527,
"grad_norm": 0.5771368508028323,
"learning_rate": 7.245972466102766e-05,
"loss": 0.6553152799606323,
"step": 640
},
{
"epoch": 0.7718242022877785,
"grad_norm": 0.6128766973307496,
"learning_rate": 7.237074466827579e-05,
"loss": 0.75093674659729,
"step": 641
},
{
"epoch": 0.7730282962071041,
"grad_norm": 0.5821257642272554,
"learning_rate": 7.228167600770101e-05,
"loss": 0.6564959287643433,
"step": 642
},
{
"epoch": 0.7742323901264299,
"grad_norm": 0.5623716196101178,
"learning_rate": 7.219251903233246e-05,
"loss": 0.8025193214416504,
"step": 643
},
{
"epoch": 0.7754364840457556,
"grad_norm": 0.5696722543963053,
"learning_rate": 7.210327409554926e-05,
"loss": 0.7683711051940918,
"step": 644
},
{
"epoch": 0.7766405779650812,
"grad_norm": 0.5976679685755757,
"learning_rate": 7.201394155107928e-05,
"loss": 0.7830510139465332,
"step": 645
},
{
"epoch": 0.777844671884407,
"grad_norm": 0.5383400096273454,
"learning_rate": 7.192452175299748e-05,
"loss": 0.7331225275993347,
"step": 646
},
{
"epoch": 0.7790487658037327,
"grad_norm": 0.5804872750816046,
"learning_rate": 7.183501505572478e-05,
"loss": 0.7439867258071899,
"step": 647
},
{
"epoch": 0.7802528597230584,
"grad_norm": 0.5709872397445639,
"learning_rate": 7.174542181402646e-05,
"loss": 0.7664347887039185,
"step": 648
},
{
"epoch": 0.7814569536423841,
"grad_norm": 0.6093539887624945,
"learning_rate": 7.165574238301085e-05,
"loss": 0.7767341136932373,
"step": 649
},
{
"epoch": 0.7826610475617098,
"grad_norm": 0.6382372128542977,
"learning_rate": 7.15659771181279e-05,
"loss": 0.780250608921051,
"step": 650
},
{
"epoch": 0.7838651414810355,
"grad_norm": 0.5817645984828969,
"learning_rate": 7.147612637516775e-05,
"loss": 0.7129443883895874,
"step": 651
},
{
"epoch": 0.7850692354003612,
"grad_norm": 0.5829365363830763,
"learning_rate": 7.138619051025935e-05,
"loss": 0.7688045501708984,
"step": 652
},
{
"epoch": 0.786273329319687,
"grad_norm": 0.5931149520853964,
"learning_rate": 7.129616987986905e-05,
"loss": 0.7989938259124756,
"step": 653
},
{
"epoch": 0.7874774232390126,
"grad_norm": 0.6079357435489189,
"learning_rate": 7.120606484079912e-05,
"loss": 0.7031389474868774,
"step": 654
},
{
"epoch": 0.7886815171583383,
"grad_norm": 0.5836757921005798,
"learning_rate": 7.111587575018648e-05,
"loss": 0.7208900451660156,
"step": 655
},
{
"epoch": 0.7898856110776641,
"grad_norm": 0.5706981336520551,
"learning_rate": 7.102560296550109e-05,
"loss": 0.6882627010345459,
"step": 656
},
{
"epoch": 0.7910897049969897,
"grad_norm": 0.5801405663826479,
"learning_rate": 7.093524684454471e-05,
"loss": 0.7287724018096924,
"step": 657
},
{
"epoch": 0.7922937989163155,
"grad_norm": 0.5112728165578423,
"learning_rate": 7.084480774544937e-05,
"loss": 0.7950488328933716,
"step": 658
},
{
"epoch": 0.7934978928356412,
"grad_norm": 0.5906716881082211,
"learning_rate": 7.075428602667602e-05,
"loss": 0.8319511413574219,
"step": 659
},
{
"epoch": 0.7947019867549668,
"grad_norm": 0.552277030288998,
"learning_rate": 7.066368204701306e-05,
"loss": 0.7356606125831604,
"step": 660
},
{
"epoch": 0.7959060806742926,
"grad_norm": 0.5491799500162513,
"learning_rate": 7.057299616557493e-05,
"loss": 0.6787431240081787,
"step": 661
},
{
"epoch": 0.7971101745936183,
"grad_norm": 0.5447685751365484,
"learning_rate": 7.048222874180072e-05,
"loss": 0.6301302313804626,
"step": 662
},
{
"epoch": 0.798314268512944,
"grad_norm": 0.5349845222628344,
"learning_rate": 7.039138013545265e-05,
"loss": 0.7335139513015747,
"step": 663
},
{
"epoch": 0.7995183624322697,
"grad_norm": 0.5308636878826329,
"learning_rate": 7.030045070661484e-05,
"loss": 0.701374888420105,
"step": 664
},
{
"epoch": 0.8007224563515954,
"grad_norm": 0.5922816809783011,
"learning_rate": 7.020944081569158e-05,
"loss": 0.7663730978965759,
"step": 665
},
{
"epoch": 0.8019265502709211,
"grad_norm": 0.5605056625652637,
"learning_rate": 7.011835082340625e-05,
"loss": 0.7345573306083679,
"step": 666
},
{
"epoch": 0.8031306441902468,
"grad_norm": 0.5851819638490449,
"learning_rate": 7.002718109079964e-05,
"loss": 0.7375038266181946,
"step": 667
},
{
"epoch": 0.8043347381095726,
"grad_norm": 0.5416749643673157,
"learning_rate": 6.993593197922852e-05,
"loss": 0.7487154603004456,
"step": 668
},
{
"epoch": 0.8055388320288982,
"grad_norm": 0.5556726879895357,
"learning_rate": 6.984460385036442e-05,
"loss": 0.7427254915237427,
"step": 669
},
{
"epoch": 0.8067429259482239,
"grad_norm": 0.6007598069149666,
"learning_rate": 6.975319706619197e-05,
"loss": 0.7750675678253174,
"step": 670
},
{
"epoch": 0.8079470198675497,
"grad_norm": 0.528992540879131,
"learning_rate": 6.966171198900761e-05,
"loss": 0.6606633067131042,
"step": 671
},
{
"epoch": 0.8091511137868754,
"grad_norm": 0.5477173596363377,
"learning_rate": 6.957014898141805e-05,
"loss": 0.8152574300765991,
"step": 672
},
{
"epoch": 0.8103552077062011,
"grad_norm": 0.6121165350927734,
"learning_rate": 6.947850840633892e-05,
"loss": 0.6906253099441528,
"step": 673
},
{
"epoch": 0.8115593016255268,
"grad_norm": 0.6828142358549254,
"learning_rate": 6.938679062699327e-05,
"loss": 0.7197479009628296,
"step": 674
},
{
"epoch": 0.8127633955448526,
"grad_norm": 0.600638365736741,
"learning_rate": 6.929499600691014e-05,
"loss": 0.7559309005737305,
"step": 675
},
{
"epoch": 0.8139674894641782,
"grad_norm": 0.5741362889435193,
"learning_rate": 6.92031249099232e-05,
"loss": 0.7413439154624939,
"step": 676
},
{
"epoch": 0.8151715833835039,
"grad_norm": 0.5573466696881405,
"learning_rate": 6.911117770016915e-05,
"loss": 0.7520703077316284,
"step": 677
},
{
"epoch": 0.8163756773028297,
"grad_norm": 0.5530340502427105,
"learning_rate": 6.901915474208644e-05,
"loss": 0.7369641661643982,
"step": 678
},
{
"epoch": 0.8175797712221553,
"grad_norm": 0.6288799148019949,
"learning_rate": 6.892705640041373e-05,
"loss": 0.6862533688545227,
"step": 679
},
{
"epoch": 0.818783865141481,
"grad_norm": 0.5705708863157596,
"learning_rate": 6.883488304018844e-05,
"loss": 0.6596237421035767,
"step": 680
},
{
"epoch": 0.8199879590608068,
"grad_norm": 0.5418277571018996,
"learning_rate": 6.874263502674538e-05,
"loss": 0.7127683758735657,
"step": 681
},
{
"epoch": 0.8211920529801324,
"grad_norm": 0.5393357810954823,
"learning_rate": 6.86503127257152e-05,
"loss": 0.6932356357574463,
"step": 682
},
{
"epoch": 0.8223961468994582,
"grad_norm": 0.5697671614974222,
"learning_rate": 6.855791650302305e-05,
"loss": 0.7497154474258423,
"step": 683
},
{
"epoch": 0.8236002408187839,
"grad_norm": 0.5306596303608581,
"learning_rate": 6.846544672488701e-05,
"loss": 0.8150767087936401,
"step": 684
},
{
"epoch": 0.8248043347381095,
"grad_norm": 0.5534085843835471,
"learning_rate": 6.837290375781678e-05,
"loss": 0.7602214217185974,
"step": 685
},
{
"epoch": 0.8260084286574353,
"grad_norm": 0.5529982450683968,
"learning_rate": 6.828028796861207e-05,
"loss": 0.7561876177787781,
"step": 686
},
{
"epoch": 0.827212522576761,
"grad_norm": 0.5484429780368382,
"learning_rate": 6.818759972436125e-05,
"loss": 0.6599973440170288,
"step": 687
},
{
"epoch": 0.8284166164960867,
"grad_norm": 0.5776866825371967,
"learning_rate": 6.809483939243992e-05,
"loss": 0.7286278009414673,
"step": 688
},
{
"epoch": 0.8296207104154124,
"grad_norm": 0.5057955934734665,
"learning_rate": 6.800200734050931e-05,
"loss": 0.7581233978271484,
"step": 689
},
{
"epoch": 0.8308248043347382,
"grad_norm": 0.5760904906160742,
"learning_rate": 6.790910393651502e-05,
"loss": 0.7653100490570068,
"step": 690
},
{
"epoch": 0.8320288982540638,
"grad_norm": 0.5984616725447457,
"learning_rate": 6.781612954868538e-05,
"loss": 0.6928480863571167,
"step": 691
},
{
"epoch": 0.8332329921733895,
"grad_norm": 0.6145944772662347,
"learning_rate": 6.77230845455301e-05,
"loss": 0.8009054660797119,
"step": 692
},
{
"epoch": 0.8344370860927153,
"grad_norm": 0.6095913953550105,
"learning_rate": 6.762996929583878e-05,
"loss": 0.6875652074813843,
"step": 693
},
{
"epoch": 0.8356411800120409,
"grad_norm": 0.5090902182729284,
"learning_rate": 6.753678416867944e-05,
"loss": 0.5892850160598755,
"step": 694
},
{
"epoch": 0.8368452739313667,
"grad_norm": 0.5661616504114457,
"learning_rate": 6.744352953339706e-05,
"loss": 0.8156208395957947,
"step": 695
},
{
"epoch": 0.8380493678506924,
"grad_norm": 0.5471603662125316,
"learning_rate": 6.735020575961213e-05,
"loss": 0.7717173099517822,
"step": 696
},
{
"epoch": 0.839253461770018,
"grad_norm": 0.5608336825556588,
"learning_rate": 6.725681321721916e-05,
"loss": 0.7841690182685852,
"step": 697
},
{
"epoch": 0.8404575556893438,
"grad_norm": 0.5652594612436186,
"learning_rate": 6.716335227638525e-05,
"loss": 0.7450107932090759,
"step": 698
},
{
"epoch": 0.8416616496086695,
"grad_norm": 0.5184526157601729,
"learning_rate": 6.706982330754858e-05,
"loss": 0.7024678587913513,
"step": 699
},
{
"epoch": 0.8428657435279951,
"grad_norm": 0.5968287295635879,
"learning_rate": 6.697622668141698e-05,
"loss": 0.774450421333313,
"step": 700
},
{
"epoch": 0.8440698374473209,
"grad_norm": 0.5615067567077281,
"learning_rate": 6.688256276896643e-05,
"loss": 0.6903250217437744,
"step": 701
},
{
"epoch": 0.8452739313666466,
"grad_norm": 0.5210615684494945,
"learning_rate": 6.678883194143962e-05,
"loss": 0.7323881387710571,
"step": 702
},
{
"epoch": 0.8464780252859723,
"grad_norm": 0.5936435413810897,
"learning_rate": 6.669503457034446e-05,
"loss": 0.7613493204116821,
"step": 703
},
{
"epoch": 0.847682119205298,
"grad_norm": 0.5477651842051788,
"learning_rate": 6.660117102745256e-05,
"loss": 0.7256999015808105,
"step": 704
},
{
"epoch": 0.8488862131246238,
"grad_norm": 0.5931823138528517,
"learning_rate": 6.650724168479789e-05,
"loss": 0.6926670670509338,
"step": 705
},
{
"epoch": 0.8500903070439494,
"grad_norm": 0.5950387212624304,
"learning_rate": 6.641324691467514e-05,
"loss": 0.7022740840911865,
"step": 706
},
{
"epoch": 0.8512944009632751,
"grad_norm": 0.5226109301283741,
"learning_rate": 6.63191870896384e-05,
"loss": 0.7763382196426392,
"step": 707
},
{
"epoch": 0.8524984948826009,
"grad_norm": 0.586021288748115,
"learning_rate": 6.622506258249956e-05,
"loss": 0.6866122484207153,
"step": 708
},
{
"epoch": 0.8537025888019265,
"grad_norm": 0.5920787573331819,
"learning_rate": 6.61308737663269e-05,
"loss": 0.6568872928619385,
"step": 709
},
{
"epoch": 0.8549066827212523,
"grad_norm": 0.5991975319472767,
"learning_rate": 6.60366210144436e-05,
"loss": 0.718787670135498,
"step": 710
},
{
"epoch": 0.856110776640578,
"grad_norm": 0.5160750999640783,
"learning_rate": 6.59423047004262e-05,
"loss": 0.7689209580421448,
"step": 711
},
{
"epoch": 0.8573148705599036,
"grad_norm": 0.5494485853850889,
"learning_rate": 6.584792519810325e-05,
"loss": 0.6794394850730896,
"step": 712
},
{
"epoch": 0.8585189644792294,
"grad_norm": 0.5219228062580127,
"learning_rate": 6.57534828815537e-05,
"loss": 0.805499792098999,
"step": 713
},
{
"epoch": 0.8597230583985551,
"grad_norm": 0.6054632002933689,
"learning_rate": 6.565897812510549e-05,
"loss": 0.6810654997825623,
"step": 714
},
{
"epoch": 0.8609271523178808,
"grad_norm": 0.5467563560047504,
"learning_rate": 6.556441130333403e-05,
"loss": 0.782832682132721,
"step": 715
},
{
"epoch": 0.8621312462372065,
"grad_norm": 0.5102699042274735,
"learning_rate": 6.546978279106074e-05,
"loss": 0.7144448757171631,
"step": 716
},
{
"epoch": 0.8633353401565322,
"grad_norm": 0.5201991055470554,
"learning_rate": 6.537509296335155e-05,
"loss": 0.7194694876670837,
"step": 717
},
{
"epoch": 0.8645394340758579,
"grad_norm": 0.576056447234189,
"learning_rate": 6.528034219551543e-05,
"loss": 0.6905584335327148,
"step": 718
},
{
"epoch": 0.8657435279951836,
"grad_norm": 0.6464466934483798,
"learning_rate": 6.518553086310285e-05,
"loss": 0.6729954481124878,
"step": 719
},
{
"epoch": 0.8669476219145094,
"grad_norm": 0.5594342791837297,
"learning_rate": 6.509065934190437e-05,
"loss": 0.8003696203231812,
"step": 720
},
{
"epoch": 0.868151715833835,
"grad_norm": 0.601593304386689,
"learning_rate": 6.499572800794911e-05,
"loss": 0.6588794589042664,
"step": 721
},
{
"epoch": 0.8693558097531607,
"grad_norm": 0.5565882109192302,
"learning_rate": 6.490073723750326e-05,
"loss": 0.7036632895469666,
"step": 722
},
{
"epoch": 0.8705599036724865,
"grad_norm": 0.5612226727107508,
"learning_rate": 6.480568740706856e-05,
"loss": 0.621507465839386,
"step": 723
},
{
"epoch": 0.8717639975918121,
"grad_norm": 0.522246530372074,
"learning_rate": 6.471057889338089e-05,
"loss": 0.6701735258102417,
"step": 724
},
{
"epoch": 0.8729680915111379,
"grad_norm": 0.5426527244446435,
"learning_rate": 6.461541207340866e-05,
"loss": 0.7284058332443237,
"step": 725
},
{
"epoch": 0.8741721854304636,
"grad_norm": 0.5593465073284523,
"learning_rate": 6.452018732435145e-05,
"loss": 0.8232830762863159,
"step": 726
},
{
"epoch": 0.8753762793497892,
"grad_norm": 0.5342600006913032,
"learning_rate": 6.442490502363838e-05,
"loss": 0.7485530972480774,
"step": 727
},
{
"epoch": 0.876580373269115,
"grad_norm": 0.5447449625195914,
"learning_rate": 6.432956554892675e-05,
"loss": 0.7344603538513184,
"step": 728
},
{
"epoch": 0.8777844671884407,
"grad_norm": 0.5322477025460601,
"learning_rate": 6.42341692781004e-05,
"loss": 0.6612800359725952,
"step": 729
},
{
"epoch": 0.8789885611077664,
"grad_norm": 0.5831757188647365,
"learning_rate": 6.413871658926833e-05,
"loss": 0.6933920979499817,
"step": 730
},
{
"epoch": 0.8801926550270921,
"grad_norm": 0.5299832721565915,
"learning_rate": 6.404320786076317e-05,
"loss": 0.7439042329788208,
"step": 731
},
{
"epoch": 0.8813967489464178,
"grad_norm": 0.5695875228926351,
"learning_rate": 6.39476434711396e-05,
"loss": 0.7039676308631897,
"step": 732
},
{
"epoch": 0.8826008428657435,
"grad_norm": 0.5832038151328962,
"learning_rate": 6.385202379917297e-05,
"loss": 0.7815375924110413,
"step": 733
},
{
"epoch": 0.8838049367850692,
"grad_norm": 0.6835928908474225,
"learning_rate": 6.375634922385775e-05,
"loss": 0.7208431959152222,
"step": 734
},
{
"epoch": 0.885009030704395,
"grad_norm": 0.6587270192090067,
"learning_rate": 6.366062012440599e-05,
"loss": 0.7941803932189941,
"step": 735
},
{
"epoch": 0.8862131246237207,
"grad_norm": 1.7245765475827708,
"learning_rate": 6.356483688024588e-05,
"loss": 0.671486496925354,
"step": 736
},
{
"epoch": 0.8874172185430463,
"grad_norm": 0.5557071279745461,
"learning_rate": 6.346899987102019e-05,
"loss": 0.7135335206985474,
"step": 737
},
{
"epoch": 0.8886213124623721,
"grad_norm": 0.4961299048808865,
"learning_rate": 6.337310947658478e-05,
"loss": 0.714313805103302,
"step": 738
},
{
"epoch": 0.8898254063816978,
"grad_norm": 0.5141981286677907,
"learning_rate": 6.327716607700719e-05,
"loss": 0.6993675231933594,
"step": 739
},
{
"epoch": 0.8910295003010235,
"grad_norm": 0.5789671720716641,
"learning_rate": 6.318117005256494e-05,
"loss": 0.7922146320343018,
"step": 740
},
{
"epoch": 0.8922335942203492,
"grad_norm": 0.5613803261760942,
"learning_rate": 6.308512178374419e-05,
"loss": 0.6571409702301025,
"step": 741
},
{
"epoch": 0.893437688139675,
"grad_norm": 0.5669966206220561,
"learning_rate": 6.298902165123815e-05,
"loss": 0.688746452331543,
"step": 742
},
{
"epoch": 0.8946417820590006,
"grad_norm": 0.5047561684826636,
"learning_rate": 6.289287003594564e-05,
"loss": 0.7370304465293884,
"step": 743
},
{
"epoch": 0.8958458759783263,
"grad_norm": 0.5563270777067858,
"learning_rate": 6.279666731896946e-05,
"loss": 0.7728006839752197,
"step": 744
},
{
"epoch": 0.8970499698976521,
"grad_norm": 0.6115646897321452,
"learning_rate": 6.270041388161503e-05,
"loss": 0.8055559992790222,
"step": 745
},
{
"epoch": 0.8982540638169777,
"grad_norm": 0.5628035739900892,
"learning_rate": 6.26041101053888e-05,
"loss": 0.7253992557525635,
"step": 746
},
{
"epoch": 0.8994581577363034,
"grad_norm": 0.5149674970238032,
"learning_rate": 6.250775637199661e-05,
"loss": 0.692879319190979,
"step": 747
},
{
"epoch": 0.9006622516556292,
"grad_norm": 0.5056342892241322,
"learning_rate": 6.241135306334254e-05,
"loss": 0.6616844534873962,
"step": 748
},
{
"epoch": 0.9018663455749548,
"grad_norm": 0.5383802540672363,
"learning_rate": 6.231490056152692e-05,
"loss": 0.6623563766479492,
"step": 749
},
{
"epoch": 0.9030704394942806,
"grad_norm": 0.5815074149661886,
"learning_rate": 6.221839924884527e-05,
"loss": 0.752768874168396,
"step": 750
},
{
"epoch": 0.9042745334136063,
"grad_norm": 0.5414073533695667,
"learning_rate": 6.21218495077864e-05,
"loss": 0.731033205986023,
"step": 751
},
{
"epoch": 0.9054786273329319,
"grad_norm": 0.5447709356541176,
"learning_rate": 6.20252517210312e-05,
"loss": 0.6803133487701416,
"step": 752
},
{
"epoch": 0.9066827212522577,
"grad_norm": 0.5378240303467464,
"learning_rate": 6.192860627145094e-05,
"loss": 0.718227744102478,
"step": 753
},
{
"epoch": 0.9078868151715834,
"grad_norm": 0.5551071834692822,
"learning_rate": 6.183191354210577e-05,
"loss": 0.7990862727165222,
"step": 754
},
{
"epoch": 0.9090909090909091,
"grad_norm": 0.5172406034430397,
"learning_rate": 6.173517391624331e-05,
"loss": 0.6609753370285034,
"step": 755
},
{
"epoch": 0.9102950030102348,
"grad_norm": 0.5736445803572223,
"learning_rate": 6.163838777729699e-05,
"loss": 0.7769580483436584,
"step": 756
},
{
"epoch": 0.9114990969295605,
"grad_norm": 0.5097735409960078,
"learning_rate": 6.154155550888466e-05,
"loss": 0.6557226181030273,
"step": 757
},
{
"epoch": 0.9127031908488862,
"grad_norm": 0.6125357429244116,
"learning_rate": 6.144467749480695e-05,
"loss": 0.7219225168228149,
"step": 758
},
{
"epoch": 0.9139072847682119,
"grad_norm": 0.5709458640554498,
"learning_rate": 6.134775411904581e-05,
"loss": 0.6767279505729675,
"step": 759
},
{
"epoch": 0.9151113786875377,
"grad_norm": 0.5556672955667038,
"learning_rate": 6.125078576576306e-05,
"loss": 0.6707606911659241,
"step": 760
},
{
"epoch": 0.9163154726068633,
"grad_norm": 0.5996100778071993,
"learning_rate": 6.115377281929867e-05,
"loss": 0.7439418435096741,
"step": 761
},
{
"epoch": 0.917519566526189,
"grad_norm": 0.6026597255285467,
"learning_rate": 6.105671566416947e-05,
"loss": 0.7168517112731934,
"step": 762
},
{
"epoch": 0.9187236604455148,
"grad_norm": 0.5205823898203639,
"learning_rate": 6.0959614685067444e-05,
"loss": 0.6558011770248413,
"step": 763
},
{
"epoch": 0.9199277543648404,
"grad_norm": 0.5171716858750717,
"learning_rate": 6.086247026685832e-05,
"loss": 0.6740440726280212,
"step": 764
},
{
"epoch": 0.9211318482841662,
"grad_norm": 0.5352969310926585,
"learning_rate": 6.0765282794579934e-05,
"loss": 0.6668114066123962,
"step": 765
},
{
"epoch": 0.9223359422034919,
"grad_norm": 0.6505808967515653,
"learning_rate": 6.066805265344084e-05,
"loss": 0.8477667570114136,
"step": 766
},
{
"epoch": 0.9235400361228175,
"grad_norm": 0.6410375604839043,
"learning_rate": 6.0570780228818705e-05,
"loss": 0.7770013213157654,
"step": 767
},
{
"epoch": 0.9247441300421433,
"grad_norm": 0.5673822560541372,
"learning_rate": 6.047346590625872e-05,
"loss": 0.7360138893127441,
"step": 768
},
{
"epoch": 0.925948223961469,
"grad_norm": 0.5557896204884362,
"learning_rate": 6.0376110071472234e-05,
"loss": 0.6802233457565308,
"step": 769
},
{
"epoch": 0.9271523178807947,
"grad_norm": 0.52113730340189,
"learning_rate": 6.027871311033506e-05,
"loss": 0.7123029828071594,
"step": 770
},
{
"epoch": 0.9283564118001204,
"grad_norm": 0.5602500741181295,
"learning_rate": 6.0181275408886064e-05,
"loss": 0.7553243637084961,
"step": 771
},
{
"epoch": 0.9295605057194462,
"grad_norm": 0.5812719342758731,
"learning_rate": 6.008379735332556e-05,
"loss": 0.8514397144317627,
"step": 772
},
{
"epoch": 0.9307645996387718,
"grad_norm": 0.5301079378776876,
"learning_rate": 5.998627933001381e-05,
"loss": 0.7405595779418945,
"step": 773
},
{
"epoch": 0.9319686935580975,
"grad_norm": 0.5524992358350457,
"learning_rate": 5.988872172546952e-05,
"loss": 0.6502660512924194,
"step": 774
},
{
"epoch": 0.9331727874774233,
"grad_norm": 0.6259792675457059,
"learning_rate": 5.979112492636824e-05,
"loss": 0.7457438111305237,
"step": 775
},
{
"epoch": 0.9343768813967489,
"grad_norm": 0.5985676283825714,
"learning_rate": 5.9693489319540906e-05,
"loss": 0.7521538734436035,
"step": 776
},
{
"epoch": 0.9355809753160746,
"grad_norm": 0.5417907461104422,
"learning_rate": 5.959581529197225e-05,
"loss": 0.6997593641281128,
"step": 777
},
{
"epoch": 0.9367850692354004,
"grad_norm": 0.5797230846074595,
"learning_rate": 5.9498103230799274e-05,
"loss": 0.802728533744812,
"step": 778
},
{
"epoch": 0.937989163154726,
"grad_norm": 0.5235298813353869,
"learning_rate": 5.940035352330975e-05,
"loss": 0.755760133266449,
"step": 779
},
{
"epoch": 0.9391932570740518,
"grad_norm": 0.5919572468116551,
"learning_rate": 5.9302566556940654e-05,
"loss": 0.7489716410636902,
"step": 780
},
{
"epoch": 0.9403973509933775,
"grad_norm": 0.5506448343636869,
"learning_rate": 5.9204742719276676e-05,
"loss": 0.6037186980247498,
"step": 781
},
{
"epoch": 0.9416014449127031,
"grad_norm": 0.565008922820836,
"learning_rate": 5.910688239804857e-05,
"loss": 0.7465704679489136,
"step": 782
},
{
"epoch": 0.9428055388320289,
"grad_norm": 0.5693013431442876,
"learning_rate": 5.900898598113177e-05,
"loss": 0.7020426392555237,
"step": 783
},
{
"epoch": 0.9440096327513546,
"grad_norm": 0.5513848988904508,
"learning_rate": 5.891105385654474e-05,
"loss": 0.6874203681945801,
"step": 784
},
{
"epoch": 0.9452137266706803,
"grad_norm": 0.5317564624539481,
"learning_rate": 5.881308641244747e-05,
"loss": 0.6125369668006897,
"step": 785
},
{
"epoch": 0.946417820590006,
"grad_norm": 0.5804792703313383,
"learning_rate": 5.871508403713997e-05,
"loss": 0.7148929238319397,
"step": 786
},
{
"epoch": 0.9476219145093318,
"grad_norm": 0.5364815284480262,
"learning_rate": 5.861704711906067e-05,
"loss": 0.7103593349456787,
"step": 787
},
{
"epoch": 0.9488260084286574,
"grad_norm": 0.6044595597833547,
"learning_rate": 5.8518976046784944e-05,
"loss": 0.6965406537055969,
"step": 788
},
{
"epoch": 0.9500301023479831,
"grad_norm": 0.5212214917819449,
"learning_rate": 5.842087120902351e-05,
"loss": 0.7151713371276855,
"step": 789
},
{
"epoch": 0.9512341962673089,
"grad_norm": 0.5559680847527447,
"learning_rate": 5.832273299462092e-05,
"loss": 0.6795516014099121,
"step": 790
},
{
"epoch": 0.9524382901866345,
"grad_norm": 0.548780921578,
"learning_rate": 5.8224561792554036e-05,
"loss": 0.7641539573669434,
"step": 791
},
{
"epoch": 0.9536423841059603,
"grad_norm": 0.5608091695330082,
"learning_rate": 5.812635799193046e-05,
"loss": 0.7201707363128662,
"step": 792
},
{
"epoch": 0.954846478025286,
"grad_norm": 0.5826010098132147,
"learning_rate": 5.802812198198699e-05,
"loss": 0.7076339721679688,
"step": 793
},
{
"epoch": 0.9560505719446116,
"grad_norm": 0.5753158331848542,
"learning_rate": 5.792985415208809e-05,
"loss": 0.709479033946991,
"step": 794
},
{
"epoch": 0.9572546658639374,
"grad_norm": 0.5566546411115479,
"learning_rate": 5.783155489172437e-05,
"loss": 0.770931601524353,
"step": 795
},
{
"epoch": 0.9584587597832631,
"grad_norm": 0.599083438626808,
"learning_rate": 5.773322459051098e-05,
"loss": 0.6527454257011414,
"step": 796
},
{
"epoch": 0.9596628537025887,
"grad_norm": 0.6290348495481346,
"learning_rate": 5.763486363818613e-05,
"loss": 0.7638934254646301,
"step": 797
},
{
"epoch": 0.9608669476219145,
"grad_norm": 0.5846705546999525,
"learning_rate": 5.75364724246095e-05,
"loss": 0.8522746562957764,
"step": 798
},
{
"epoch": 0.9620710415412402,
"grad_norm": 0.5249812879164455,
"learning_rate": 5.743805133976071e-05,
"loss": 0.7056288123130798,
"step": 799
},
{
"epoch": 0.963275135460566,
"grad_norm": 0.5870806629180437,
"learning_rate": 5.733960077373781e-05,
"loss": 0.7362521886825562,
"step": 800
},
{
"epoch": 0.9644792293798916,
"grad_norm": 0.5992634913094184,
"learning_rate": 5.7241121116755646e-05,
"loss": 0.7288447618484497,
"step": 801
},
{
"epoch": 0.9656833232992174,
"grad_norm": 0.5469756167657702,
"learning_rate": 5.714261275914442e-05,
"loss": 0.7934062480926514,
"step": 802
},
{
"epoch": 0.9668874172185431,
"grad_norm": 0.6496356256264952,
"learning_rate": 5.704407609134805e-05,
"loss": 0.8269863724708557,
"step": 803
},
{
"epoch": 0.9680915111378687,
"grad_norm": 0.5037971848741302,
"learning_rate": 5.694551150392271e-05,
"loss": 0.7017319202423096,
"step": 804
},
{
"epoch": 0.9692956050571945,
"grad_norm": 0.5581416866027927,
"learning_rate": 5.684691938753517e-05,
"loss": 0.6454797387123108,
"step": 805
},
{
"epoch": 0.9704996989765202,
"grad_norm": 0.5226184565373623,
"learning_rate": 5.674830013296137e-05,
"loss": 0.6731082201004028,
"step": 806
},
{
"epoch": 0.9717037928958459,
"grad_norm": 0.5182166828064974,
"learning_rate": 5.664965413108481e-05,
"loss": 0.6522151231765747,
"step": 807
},
{
"epoch": 0.9729078868151716,
"grad_norm": 0.5685770883500431,
"learning_rate": 5.655098177289496e-05,
"loss": 0.7161366939544678,
"step": 808
},
{
"epoch": 0.9741119807344973,
"grad_norm": 0.5555404388639458,
"learning_rate": 5.6452283449485774e-05,
"loss": 0.7227890491485596,
"step": 809
},
{
"epoch": 0.975316074653823,
"grad_norm": 0.5622492078354654,
"learning_rate": 5.635355955205416e-05,
"loss": 0.7163590788841248,
"step": 810
},
{
"epoch": 0.9765201685731487,
"grad_norm": 0.4986516512531464,
"learning_rate": 5.625481047189835e-05,
"loss": 0.5898875594139099,
"step": 811
},
{
"epoch": 0.9777242624924745,
"grad_norm": 0.5559389458013995,
"learning_rate": 5.6156036600416385e-05,
"loss": 0.644203245639801,
"step": 812
},
{
"epoch": 0.9789283564118001,
"grad_norm": 0.5577887920278128,
"learning_rate": 5.60572383291046e-05,
"loss": 0.7715066075325012,
"step": 813
},
{
"epoch": 0.9801324503311258,
"grad_norm": 0.5694954155998118,
"learning_rate": 5.595841604955601e-05,
"loss": 0.7114702463150024,
"step": 814
},
{
"epoch": 0.9813365442504516,
"grad_norm": 0.5728384268780161,
"learning_rate": 5.585957015345882e-05,
"loss": 0.7500208020210266,
"step": 815
},
{
"epoch": 0.9825406381697772,
"grad_norm": 0.5754460458927252,
"learning_rate": 5.5760701032594775e-05,
"loss": 0.8016777634620667,
"step": 816
},
{
"epoch": 0.983744732089103,
"grad_norm": 0.5378230131701004,
"learning_rate": 5.566180907883777e-05,
"loss": 0.6485020518302917,
"step": 817
},
{
"epoch": 0.9849488260084287,
"grad_norm": 0.6119817727352167,
"learning_rate": 5.556289468415212e-05,
"loss": 0.7693502306938171,
"step": 818
},
{
"epoch": 0.9861529199277543,
"grad_norm": 0.5318784574757923,
"learning_rate": 5.546395824059113e-05,
"loss": 0.7719030380249023,
"step": 819
},
{
"epoch": 0.9873570138470801,
"grad_norm": 0.5867784164801246,
"learning_rate": 5.536500014029547e-05,
"loss": 0.7008310556411743,
"step": 820
},
{
"epoch": 0.9885611077664058,
"grad_norm": 0.5069862861453538,
"learning_rate": 5.5266020775491654e-05,
"loss": 0.6411416530609131,
"step": 821
},
{
"epoch": 0.9897652016857315,
"grad_norm": 0.5834044578675054,
"learning_rate": 5.5167020538490514e-05,
"loss": 0.717340350151062,
"step": 822
},
{
"epoch": 0.9909692956050572,
"grad_norm": 0.5379644400350362,
"learning_rate": 5.506799982168553e-05,
"loss": 0.6762273907661438,
"step": 823
},
{
"epoch": 0.9921733895243829,
"grad_norm": 0.529911656567608,
"learning_rate": 5.496895901755145e-05,
"loss": 0.7467926740646362,
"step": 824
},
{
"epoch": 0.9933774834437086,
"grad_norm": 0.5518942402789531,
"learning_rate": 5.486989851864258e-05,
"loss": 0.7194281220436096,
"step": 825
},
{
"epoch": 0.9945815773630343,
"grad_norm": 0.5498490180172129,
"learning_rate": 5.47708187175913e-05,
"loss": 0.7484912872314453,
"step": 826
},
{
"epoch": 0.9957856712823601,
"grad_norm": 0.5493140404581905,
"learning_rate": 5.4671720007106507e-05,
"loss": 0.667789876461029,
"step": 827
},
{
"epoch": 0.9969897652016857,
"grad_norm": 0.5254892526478596,
"learning_rate": 5.4572602779972006e-05,
"loss": 0.6494565606117249,
"step": 828
},
{
"epoch": 0.9981938591210114,
"grad_norm": 0.5252729042272404,
"learning_rate": 5.447346742904508e-05,
"loss": 0.6520144939422607,
"step": 829
},
{
"epoch": 0.9993979530403372,
"grad_norm": 0.5079551233357467,
"learning_rate": 5.437431434725473e-05,
"loss": 0.7297295928001404,
"step": 830
},
{
"epoch": 1.0,
"grad_norm": 0.7548489418432899,
"learning_rate": 5.427514392760034e-05,
"loss": 0.7330588102340698,
"step": 831
},
{
"epoch": 1.0012040939193256,
"grad_norm": 0.48262108535450576,
"learning_rate": 5.417595656314997e-05,
"loss": 0.6693322658538818,
"step": 832
},
{
"epoch": 1.0024081878386515,
"grad_norm": 0.5473512895400998,
"learning_rate": 5.40767526470388e-05,
"loss": 0.577499270439148,
"step": 833
},
{
"epoch": 1.0036122817579771,
"grad_norm": 0.5567998652776117,
"learning_rate": 5.39775325724677e-05,
"loss": 0.7308440208435059,
"step": 834
},
{
"epoch": 1.0048163756773028,
"grad_norm": 0.4710834471702117,
"learning_rate": 5.3878296732701515e-05,
"loss": 0.5792309045791626,
"step": 835
},
{
"epoch": 1.0060204695966286,
"grad_norm": 0.499021670786511,
"learning_rate": 5.377904552106763e-05,
"loss": 0.5693698525428772,
"step": 836
},
{
"epoch": 1.0072245635159542,
"grad_norm": 0.5243691549099818,
"learning_rate": 5.367977933095428e-05,
"loss": 0.7172250747680664,
"step": 837
},
{
"epoch": 1.0084286574352799,
"grad_norm": 0.5110304762735773,
"learning_rate": 5.3580498555809163e-05,
"loss": 0.6046251654624939,
"step": 838
},
{
"epoch": 1.0096327513546057,
"grad_norm": 0.5114687722760671,
"learning_rate": 5.348120358913773e-05,
"loss": 0.6995025277137756,
"step": 839
},
{
"epoch": 1.0108368452739314,
"grad_norm": 0.5912572180109141,
"learning_rate": 5.338189482450167e-05,
"loss": 0.616209089756012,
"step": 840
},
{
"epoch": 1.012040939193257,
"grad_norm": 0.5140400551380612,
"learning_rate": 5.3282572655517416e-05,
"loss": 0.5996021628379822,
"step": 841
},
{
"epoch": 1.0132450331125828,
"grad_norm": 0.5391679793572282,
"learning_rate": 5.318323747585444e-05,
"loss": 0.7038933038711548,
"step": 842
},
{
"epoch": 1.0144491270319085,
"grad_norm": 0.5193952980537593,
"learning_rate": 5.308388967923391e-05,
"loss": 0.5445797443389893,
"step": 843
},
{
"epoch": 1.0156532209512341,
"grad_norm": 0.5620145760135244,
"learning_rate": 5.298452965942687e-05,
"loss": 0.5499206781387329,
"step": 844
},
{
"epoch": 1.01685731487056,
"grad_norm": 0.5390761246484093,
"learning_rate": 5.2885157810252915e-05,
"loss": 0.4958648085594177,
"step": 845
},
{
"epoch": 1.0180614087898856,
"grad_norm": 0.5925470805964254,
"learning_rate": 5.278577452557845e-05,
"loss": 0.6143114566802979,
"step": 846
},
{
"epoch": 1.0192655027092112,
"grad_norm": 0.5570967510117002,
"learning_rate": 5.2686380199315244e-05,
"loss": 0.6284155249595642,
"step": 847
},
{
"epoch": 1.020469596628537,
"grad_norm": 0.5829210418808195,
"learning_rate": 5.2586975225418854e-05,
"loss": 0.6126077175140381,
"step": 848
},
{
"epoch": 1.0216736905478627,
"grad_norm": 0.663542970695612,
"learning_rate": 5.248755999788699e-05,
"loss": 0.5881949663162231,
"step": 849
},
{
"epoch": 1.0228777844671884,
"grad_norm": 0.508265452000293,
"learning_rate": 5.2388134910758015e-05,
"loss": 0.49111035466194153,
"step": 850
},
{
"epoch": 1.0240818783865142,
"grad_norm": 0.5947010917583964,
"learning_rate": 5.2288700358109375e-05,
"loss": 0.5964053869247437,
"step": 851
},
{
"epoch": 1.0252859723058398,
"grad_norm": 0.6478106165020875,
"learning_rate": 5.218925673405607e-05,
"loss": 0.5837634205818176,
"step": 852
},
{
"epoch": 1.0264900662251655,
"grad_norm": 0.5566597298872943,
"learning_rate": 5.208980443274899e-05,
"loss": 0.6266411542892456,
"step": 853
},
{
"epoch": 1.0276941601444913,
"grad_norm": 0.6119087058605345,
"learning_rate": 5.199034384837345e-05,
"loss": 0.5852352380752563,
"step": 854
},
{
"epoch": 1.028898254063817,
"grad_norm": 0.5803055714357055,
"learning_rate": 5.189087537514763e-05,
"loss": 0.6915392875671387,
"step": 855
},
{
"epoch": 1.0301023479831426,
"grad_norm": 0.667773625926801,
"learning_rate": 5.179139940732091e-05,
"loss": 0.6426236629486084,
"step": 856
},
{
"epoch": 1.0313064419024685,
"grad_norm": 0.6857940311337501,
"learning_rate": 5.169191633917242e-05,
"loss": 0.6479716300964355,
"step": 857
},
{
"epoch": 1.032510535821794,
"grad_norm": 0.5987681942027812,
"learning_rate": 5.159242656500942e-05,
"loss": 0.5855128765106201,
"step": 858
},
{
"epoch": 1.0337146297411197,
"grad_norm": 0.6431412742256677,
"learning_rate": 5.149293047916576e-05,
"loss": 0.6104673743247986,
"step": 859
},
{
"epoch": 1.0349187236604456,
"grad_norm": 0.6469125810000602,
"learning_rate": 5.139342847600028e-05,
"loss": 0.5815907120704651,
"step": 860
},
{
"epoch": 1.0361228175797712,
"grad_norm": 0.6322944781362547,
"learning_rate": 5.12939209498953e-05,
"loss": 0.6261488199234009,
"step": 861
},
{
"epoch": 1.0373269114990968,
"grad_norm": 0.5933004058886241,
"learning_rate": 5.119440829525504e-05,
"loss": 0.5986544489860535,
"step": 862
},
{
"epoch": 1.0385310054184227,
"grad_norm": 0.6017294433054711,
"learning_rate": 5.1094890906504e-05,
"loss": 0.5497260689735413,
"step": 863
},
{
"epoch": 1.0397350993377483,
"grad_norm": 0.6545529261884225,
"learning_rate": 5.0995369178085484e-05,
"loss": 0.6652652621269226,
"step": 864
},
{
"epoch": 1.040939193257074,
"grad_norm": 0.5884144368772644,
"learning_rate": 5.0895843504460005e-05,
"loss": 0.5726611614227295,
"step": 865
},
{
"epoch": 1.0421432871763998,
"grad_norm": 0.612888846785316,
"learning_rate": 5.0796314280103664e-05,
"loss": 0.6019963026046753,
"step": 866
},
{
"epoch": 1.0433473810957254,
"grad_norm": 0.6341773077017514,
"learning_rate": 5.0696781899506686e-05,
"loss": 0.6352479457855225,
"step": 867
},
{
"epoch": 1.044551475015051,
"grad_norm": 0.6050053156332715,
"learning_rate": 5.059724675717177e-05,
"loss": 0.6951148509979248,
"step": 868
},
{
"epoch": 1.045755568934377,
"grad_norm": 0.6092115250909115,
"learning_rate": 5.049770924761259e-05,
"loss": 0.609517514705658,
"step": 869
},
{
"epoch": 1.0469596628537026,
"grad_norm": 0.6093665118132492,
"learning_rate": 5.039816976535219e-05,
"loss": 0.6204075813293457,
"step": 870
},
{
"epoch": 1.0481637567730282,
"grad_norm": 0.5416011274391686,
"learning_rate": 5.029862870492142e-05,
"loss": 0.5534996390342712,
"step": 871
},
{
"epoch": 1.049367850692354,
"grad_norm": 0.5815173865104868,
"learning_rate": 5.0199086460857406e-05,
"loss": 0.6000134944915771,
"step": 872
},
{
"epoch": 1.0505719446116797,
"grad_norm": 0.6149736209209395,
"learning_rate": 5.0099543427701956e-05,
"loss": 0.6076500415802002,
"step": 873
},
{
"epoch": 1.0517760385310053,
"grad_norm": 0.6287626380440837,
"learning_rate": 5e-05,
"loss": 0.6155140399932861,
"step": 874
},
{
"epoch": 1.0529801324503312,
"grad_norm": 0.5717481586151558,
"learning_rate": 4.9900456572298055e-05,
"loss": 0.5186606645584106,
"step": 875
},
{
"epoch": 1.0541842263696568,
"grad_norm": 0.6226925894026947,
"learning_rate": 4.980091353914259e-05,
"loss": 0.6474608778953552,
"step": 876
},
{
"epoch": 1.0553883202889824,
"grad_norm": 0.6964903156335773,
"learning_rate": 4.9701371295078603e-05,
"loss": 0.6780328154563904,
"step": 877
},
{
"epoch": 1.0565924142083083,
"grad_norm": 0.5756463569010563,
"learning_rate": 4.9601830234647824e-05,
"loss": 0.6036153435707092,
"step": 878
},
{
"epoch": 1.057796508127634,
"grad_norm": 0.6520552306333011,
"learning_rate": 4.950229075238742e-05,
"loss": 0.6786141395568848,
"step": 879
},
{
"epoch": 1.0590006020469596,
"grad_norm": 0.5848176085725552,
"learning_rate": 4.940275324282824e-05,
"loss": 0.5802556872367859,
"step": 880
},
{
"epoch": 1.0602046959662854,
"grad_norm": 0.6222612587675024,
"learning_rate": 4.930321810049334e-05,
"loss": 0.551266074180603,
"step": 881
},
{
"epoch": 1.061408789885611,
"grad_norm": 0.6335949011450209,
"learning_rate": 4.920368571989636e-05,
"loss": 0.5886940956115723,
"step": 882
},
{
"epoch": 1.062612883804937,
"grad_norm": 0.5668099378869348,
"learning_rate": 4.910415649554001e-05,
"loss": 0.6376104950904846,
"step": 883
},
{
"epoch": 1.0638169777242625,
"grad_norm": 0.6343326763172722,
"learning_rate": 4.900463082191452e-05,
"loss": 0.5883640050888062,
"step": 884
},
{
"epoch": 1.0650210716435882,
"grad_norm": 0.6120672237185096,
"learning_rate": 4.890510909349602e-05,
"loss": 0.5998228788375854,
"step": 885
},
{
"epoch": 1.0662251655629138,
"grad_norm": 0.6466629735077495,
"learning_rate": 4.880559170474499e-05,
"loss": 0.6147492527961731,
"step": 886
},
{
"epoch": 1.0674292594822397,
"grad_norm": 0.6188477863366557,
"learning_rate": 4.870607905010471e-05,
"loss": 0.5260412096977234,
"step": 887
},
{
"epoch": 1.0686333534015653,
"grad_norm": 0.6531014500952979,
"learning_rate": 4.860657152399973e-05,
"loss": 0.6408224701881409,
"step": 888
},
{
"epoch": 1.0698374473208911,
"grad_norm": 0.6356448658174779,
"learning_rate": 4.850706952083426e-05,
"loss": 0.6420581340789795,
"step": 889
},
{
"epoch": 1.0710415412402168,
"grad_norm": 0.6638429959160749,
"learning_rate": 4.840757343499059e-05,
"loss": 0.5597264766693115,
"step": 890
},
{
"epoch": 1.0722456351595424,
"grad_norm": 0.6316267115529358,
"learning_rate": 4.83080836608276e-05,
"loss": 0.5977203845977783,
"step": 891
},
{
"epoch": 1.073449729078868,
"grad_norm": 0.6332312497061402,
"learning_rate": 4.82086005926791e-05,
"loss": 0.5730206370353699,
"step": 892
},
{
"epoch": 1.074653822998194,
"grad_norm": 0.6036387570021288,
"learning_rate": 4.8109124624852386e-05,
"loss": 0.5876260995864868,
"step": 893
},
{
"epoch": 1.0758579169175195,
"grad_norm": 0.6209977622958593,
"learning_rate": 4.800965615162655e-05,
"loss": 0.6396963596343994,
"step": 894
},
{
"epoch": 1.0770620108368454,
"grad_norm": 0.661525637487104,
"learning_rate": 4.791019556725104e-05,
"loss": 0.5814670920372009,
"step": 895
},
{
"epoch": 1.078266104756171,
"grad_norm": 0.6275103232059692,
"learning_rate": 4.7810743265943955e-05,
"loss": 0.6844623684883118,
"step": 896
},
{
"epoch": 1.0794701986754967,
"grad_norm": 0.6818646104901908,
"learning_rate": 4.771129964189063e-05,
"loss": 0.6540423631668091,
"step": 897
},
{
"epoch": 1.0806742925948223,
"grad_norm": 0.6899606024380505,
"learning_rate": 4.7611865089242004e-05,
"loss": 0.6030701398849487,
"step": 898
},
{
"epoch": 1.0818783865141481,
"grad_norm": 0.6156599517777511,
"learning_rate": 4.751244000211302e-05,
"loss": 0.6412866711616516,
"step": 899
},
{
"epoch": 1.0830824804334738,
"grad_norm": 0.5724892187762796,
"learning_rate": 4.741302477458116e-05,
"loss": 0.5549468398094177,
"step": 900
},
{
"epoch": 1.0842865743527996,
"grad_norm": 0.6770803099451937,
"learning_rate": 4.731361980068476e-05,
"loss": 0.5970389246940613,
"step": 901
},
{
"epoch": 1.0854906682721253,
"grad_norm": 0.6628879473222443,
"learning_rate": 4.7214225474421556e-05,
"loss": 0.6336625814437866,
"step": 902
},
{
"epoch": 1.086694762191451,
"grad_norm": 0.7118808898216501,
"learning_rate": 4.7114842189747096e-05,
"loss": 0.5842564702033997,
"step": 903
},
{
"epoch": 1.0878988561107765,
"grad_norm": 0.6808325766019172,
"learning_rate": 4.701547034057313e-05,
"loss": 0.6561811566352844,
"step": 904
},
{
"epoch": 1.0891029500301024,
"grad_norm": 0.6883360456461192,
"learning_rate": 4.691611032076611e-05,
"loss": 0.5866090059280396,
"step": 905
},
{
"epoch": 1.090307043949428,
"grad_norm": 0.594597876069142,
"learning_rate": 4.6816762524145565e-05,
"loss": 0.5930664539337158,
"step": 906
},
{
"epoch": 1.0915111378687539,
"grad_norm": 0.6540482900479212,
"learning_rate": 4.67174273444826e-05,
"loss": 0.6051517724990845,
"step": 907
},
{
"epoch": 1.0927152317880795,
"grad_norm": 0.6350023068043041,
"learning_rate": 4.6618105175498336e-05,
"loss": 0.5787694454193115,
"step": 908
},
{
"epoch": 1.0939193257074051,
"grad_norm": 0.6003852495110125,
"learning_rate": 4.6518796410862284e-05,
"loss": 0.474132776260376,
"step": 909
},
{
"epoch": 1.095123419626731,
"grad_norm": 0.6643572746823975,
"learning_rate": 4.641950144419085e-05,
"loss": 0.5836554765701294,
"step": 910
},
{
"epoch": 1.0963275135460566,
"grad_norm": 0.6346538134124894,
"learning_rate": 4.632022066904573e-05,
"loss": 0.632318913936615,
"step": 911
},
{
"epoch": 1.0975316074653823,
"grad_norm": 0.6506542788169488,
"learning_rate": 4.622095447893238e-05,
"loss": 0.6343862414360046,
"step": 912
},
{
"epoch": 1.098735701384708,
"grad_norm": 0.7055409962557284,
"learning_rate": 4.612170326729849e-05,
"loss": 0.5320890545845032,
"step": 913
},
{
"epoch": 1.0999397953040337,
"grad_norm": 0.6710841433347294,
"learning_rate": 4.6022467427532316e-05,
"loss": 0.6487467288970947,
"step": 914
},
{
"epoch": 1.1011438892233594,
"grad_norm": 0.6161335106985909,
"learning_rate": 4.592324735296122e-05,
"loss": 0.5364488363265991,
"step": 915
},
{
"epoch": 1.1023479831426852,
"grad_norm": 0.6567852610772589,
"learning_rate": 4.582404343685005e-05,
"loss": 0.5512426495552063,
"step": 916
},
{
"epoch": 1.1035520770620109,
"grad_norm": 0.7683181606199703,
"learning_rate": 4.5724856072399666e-05,
"loss": 0.6094173192977905,
"step": 917
},
{
"epoch": 1.1047561709813365,
"grad_norm": 0.6291985708032748,
"learning_rate": 4.562568565274528e-05,
"loss": 0.5325632691383362,
"step": 918
},
{
"epoch": 1.1059602649006623,
"grad_norm": 0.6285445914420223,
"learning_rate": 4.552653257095495e-05,
"loss": 0.6411327123641968,
"step": 919
},
{
"epoch": 1.107164358819988,
"grad_norm": 0.6778912339737762,
"learning_rate": 4.5427397220028006e-05,
"loss": 0.6963979005813599,
"step": 920
},
{
"epoch": 1.1083684527393136,
"grad_norm": 0.660219208281244,
"learning_rate": 4.5328279992893505e-05,
"loss": 0.5509607195854187,
"step": 921
},
{
"epoch": 1.1095725466586395,
"grad_norm": 0.5881472136676991,
"learning_rate": 4.5229181282408705e-05,
"loss": 0.5656740069389343,
"step": 922
},
{
"epoch": 1.110776640577965,
"grad_norm": 0.6285472375245976,
"learning_rate": 4.513010148135743e-05,
"loss": 0.5325478315353394,
"step": 923
},
{
"epoch": 1.1119807344972907,
"grad_norm": 0.6573263156024722,
"learning_rate": 4.5031040982448564e-05,
"loss": 0.568873405456543,
"step": 924
},
{
"epoch": 1.1131848284166166,
"grad_norm": 0.6375892634033876,
"learning_rate": 4.493200017831448e-05,
"loss": 0.5534123182296753,
"step": 925
},
{
"epoch": 1.1143889223359422,
"grad_norm": 0.6047252869641281,
"learning_rate": 4.48329794615095e-05,
"loss": 0.6317431330680847,
"step": 926
},
{
"epoch": 1.1155930162552679,
"grad_norm": 0.7466768979825729,
"learning_rate": 4.4733979224508344e-05,
"loss": 0.6036220192909241,
"step": 927
},
{
"epoch": 1.1167971101745937,
"grad_norm": 0.6792290210916058,
"learning_rate": 4.4634999859704546e-05,
"loss": 0.6010956764221191,
"step": 928
},
{
"epoch": 1.1180012040939193,
"grad_norm": 0.63632818802347,
"learning_rate": 4.453604175940888e-05,
"loss": 0.6214673519134521,
"step": 929
},
{
"epoch": 1.119205298013245,
"grad_norm": 0.5890975902667656,
"learning_rate": 4.443710531584789e-05,
"loss": 0.5909805297851562,
"step": 930
},
{
"epoch": 1.1204093919325708,
"grad_norm": 0.7758035748650709,
"learning_rate": 4.433819092116223e-05,
"loss": 0.6618773937225342,
"step": 931
},
{
"epoch": 1.1216134858518965,
"grad_norm": 0.6032633893730255,
"learning_rate": 4.423929896740522e-05,
"loss": 0.6046672463417053,
"step": 932
},
{
"epoch": 1.122817579771222,
"grad_norm": 0.5963867825539027,
"learning_rate": 4.41404298465412e-05,
"loss": 0.557878851890564,
"step": 933
},
{
"epoch": 1.124021673690548,
"grad_norm": 0.6278341782515403,
"learning_rate": 4.4041583950444004e-05,
"loss": 0.5768659710884094,
"step": 934
},
{
"epoch": 1.1252257676098736,
"grad_norm": 0.6692340070481475,
"learning_rate": 4.3942761670895416e-05,
"loss": 0.596519410610199,
"step": 935
},
{
"epoch": 1.1264298615291992,
"grad_norm": 0.6324155300577929,
"learning_rate": 4.384396339958361e-05,
"loss": 0.6091715097427368,
"step": 936
},
{
"epoch": 1.127633955448525,
"grad_norm": 0.6873308697173252,
"learning_rate": 4.374518952810167e-05,
"loss": 0.6679472923278809,
"step": 937
},
{
"epoch": 1.1288380493678507,
"grad_norm": 0.60347781133788,
"learning_rate": 4.364644044794585e-05,
"loss": 0.5072171688079834,
"step": 938
},
{
"epoch": 1.1300421432871763,
"grad_norm": 0.5958005061206001,
"learning_rate": 4.354771655051424e-05,
"loss": 0.45150619745254517,
"step": 939
},
{
"epoch": 1.1312462372065022,
"grad_norm": 0.6458018431543086,
"learning_rate": 4.344901822710506e-05,
"loss": 0.5927637815475464,
"step": 940
},
{
"epoch": 1.1324503311258278,
"grad_norm": 0.6299802211614689,
"learning_rate": 4.33503458689152e-05,
"loss": 0.6127684712409973,
"step": 941
},
{
"epoch": 1.1336544250451535,
"grad_norm": 0.6064791911526438,
"learning_rate": 4.3251699867038634e-05,
"loss": 0.5717106461524963,
"step": 942
},
{
"epoch": 1.1348585189644793,
"grad_norm": 0.5664911073371133,
"learning_rate": 4.3153080612464835e-05,
"loss": 0.5828589200973511,
"step": 943
},
{
"epoch": 1.136062612883805,
"grad_norm": 0.6168659838027578,
"learning_rate": 4.305448849607731e-05,
"loss": 0.6533054113388062,
"step": 944
},
{
"epoch": 1.1372667068031306,
"grad_norm": 0.6620892379351815,
"learning_rate": 4.295592390865194e-05,
"loss": 0.625812292098999,
"step": 945
},
{
"epoch": 1.1384708007224564,
"grad_norm": 0.627104880201231,
"learning_rate": 4.28573872408556e-05,
"loss": 0.6203385591506958,
"step": 946
},
{
"epoch": 1.139674894641782,
"grad_norm": 0.6520709720872566,
"learning_rate": 4.2758878883244366e-05,
"loss": 0.6454954147338867,
"step": 947
},
{
"epoch": 1.1408789885611077,
"grad_norm": 0.5670002197473883,
"learning_rate": 4.2660399226262205e-05,
"loss": 0.6776309013366699,
"step": 948
},
{
"epoch": 1.1420830824804336,
"grad_norm": 0.6651897379271099,
"learning_rate": 4.256194866023929e-05,
"loss": 0.6165192127227783,
"step": 949
},
{
"epoch": 1.1432871763997592,
"grad_norm": 0.6070318758880671,
"learning_rate": 4.24635275753905e-05,
"loss": 0.6247190833091736,
"step": 950
},
{
"epoch": 1.1444912703190848,
"grad_norm": 0.6594245265551396,
"learning_rate": 4.236513636181389e-05,
"loss": 0.6294478178024292,
"step": 951
},
{
"epoch": 1.1456953642384107,
"grad_norm": 0.6656388910182744,
"learning_rate": 4.2266775409489023e-05,
"loss": 0.6407076716423035,
"step": 952
},
{
"epoch": 1.1468994581577363,
"grad_norm": 0.6001528475820775,
"learning_rate": 4.2168445108275636e-05,
"loss": 0.5295635461807251,
"step": 953
},
{
"epoch": 1.148103552077062,
"grad_norm": 0.6623205303719578,
"learning_rate": 4.2070145847911915e-05,
"loss": 0.6363142132759094,
"step": 954
},
{
"epoch": 1.1493076459963878,
"grad_norm": 0.6595593517698526,
"learning_rate": 4.197187801801301e-05,
"loss": 0.5908467769622803,
"step": 955
},
{
"epoch": 1.1505117399157134,
"grad_norm": 0.6416992225837528,
"learning_rate": 4.187364200806956e-05,
"loss": 0.6078401803970337,
"step": 956
},
{
"epoch": 1.151715833835039,
"grad_norm": 0.6426590726024174,
"learning_rate": 4.177543820744597e-05,
"loss": 0.5659914016723633,
"step": 957
},
{
"epoch": 1.152919927754365,
"grad_norm": 0.5945058633012985,
"learning_rate": 4.167726700537909e-05,
"loss": 0.5272383689880371,
"step": 958
},
{
"epoch": 1.1541240216736905,
"grad_norm": 0.6517936478039313,
"learning_rate": 4.157912879097651e-05,
"loss": 0.6436958909034729,
"step": 959
},
{
"epoch": 1.1553281155930162,
"grad_norm": 0.6378579002832153,
"learning_rate": 4.1481023953215074e-05,
"loss": 0.6222982406616211,
"step": 960
},
{
"epoch": 1.156532209512342,
"grad_norm": 0.6050181971864277,
"learning_rate": 4.1382952880939346e-05,
"loss": 0.525058925151825,
"step": 961
},
{
"epoch": 1.1577363034316677,
"grad_norm": 0.6211097246331354,
"learning_rate": 4.128491596286004e-05,
"loss": 0.559781014919281,
"step": 962
},
{
"epoch": 1.1589403973509933,
"grad_norm": 0.6174026620410265,
"learning_rate": 4.118691358755254e-05,
"loss": 0.5967716574668884,
"step": 963
},
{
"epoch": 1.1601444912703192,
"grad_norm": 0.665933154143997,
"learning_rate": 4.108894614345527e-05,
"loss": 0.6097081303596497,
"step": 964
},
{
"epoch": 1.1613485851896448,
"grad_norm": 0.7893212528385332,
"learning_rate": 4.0991014018868255e-05,
"loss": 0.6505680084228516,
"step": 965
},
{
"epoch": 1.1625526791089704,
"grad_norm": 0.6246966439954491,
"learning_rate": 4.0893117601951434e-05,
"loss": 0.5621920228004456,
"step": 966
},
{
"epoch": 1.1637567730282963,
"grad_norm": 0.8222739403785936,
"learning_rate": 4.079525728072334e-05,
"loss": 0.5986034274101257,
"step": 967
},
{
"epoch": 1.164960866947622,
"grad_norm": 0.7050682371518708,
"learning_rate": 4.0697433443059344e-05,
"loss": 0.5873398780822754,
"step": 968
},
{
"epoch": 1.1661649608669475,
"grad_norm": 0.5796676608087229,
"learning_rate": 4.059964647669025e-05,
"loss": 0.5936042070388794,
"step": 969
},
{
"epoch": 1.1673690547862734,
"grad_norm": 0.787700686593201,
"learning_rate": 4.050189676920075e-05,
"loss": 0.5301788449287415,
"step": 970
},
{
"epoch": 1.168573148705599,
"grad_norm": 0.6443250874812158,
"learning_rate": 4.0404184708027764e-05,
"loss": 0.6244308948516846,
"step": 971
},
{
"epoch": 1.1697772426249247,
"grad_norm": 0.6420455168036301,
"learning_rate": 4.03065106804591e-05,
"loss": 0.5246211290359497,
"step": 972
},
{
"epoch": 1.1709813365442505,
"grad_norm": 0.6233558947698492,
"learning_rate": 4.0208875073631767e-05,
"loss": 0.583974301815033,
"step": 973
},
{
"epoch": 1.1721854304635762,
"grad_norm": 0.6674639893979444,
"learning_rate": 4.01112782745305e-05,
"loss": 0.6854992508888245,
"step": 974
},
{
"epoch": 1.1733895243829018,
"grad_norm": 0.6761020516050178,
"learning_rate": 4.001372066998621e-05,
"loss": 0.47961360216140747,
"step": 975
},
{
"epoch": 1.1745936183022276,
"grad_norm": 0.6937061007497107,
"learning_rate": 3.9916202646674454e-05,
"loss": 0.5944783091545105,
"step": 976
},
{
"epoch": 1.1757977122215533,
"grad_norm": 0.7056015316208057,
"learning_rate": 3.981872459111394e-05,
"loss": 0.5598882436752319,
"step": 977
},
{
"epoch": 1.177001806140879,
"grad_norm": 0.6404405450738602,
"learning_rate": 3.9721286889664946e-05,
"loss": 0.576660692691803,
"step": 978
},
{
"epoch": 1.1782059000602048,
"grad_norm": 0.7104592496085945,
"learning_rate": 3.962388992852778e-05,
"loss": 0.6423761248588562,
"step": 979
},
{
"epoch": 1.1794099939795304,
"grad_norm": 0.6594542729046524,
"learning_rate": 3.9526534093741294e-05,
"loss": 0.629199743270874,
"step": 980
},
{
"epoch": 1.180614087898856,
"grad_norm": 0.6630058301620763,
"learning_rate": 3.94292197711813e-05,
"loss": 0.5926947593688965,
"step": 981
},
{
"epoch": 1.1818181818181819,
"grad_norm": 0.6813208402691311,
"learning_rate": 3.933194734655916e-05,
"loss": 0.6430405378341675,
"step": 982
},
{
"epoch": 1.1830222757375075,
"grad_norm": 0.6509982974967221,
"learning_rate": 3.923471720542007e-05,
"loss": 0.6401696801185608,
"step": 983
},
{
"epoch": 1.1842263696568331,
"grad_norm": 0.6536388401019356,
"learning_rate": 3.9137529733141705e-05,
"loss": 0.6143403053283691,
"step": 984
},
{
"epoch": 1.185430463576159,
"grad_norm": 0.714504675861702,
"learning_rate": 3.904038531493257e-05,
"loss": 0.7203058004379272,
"step": 985
},
{
"epoch": 1.1866345574954846,
"grad_norm": 0.6173945491470922,
"learning_rate": 3.894328433583053e-05,
"loss": 0.6278570294380188,
"step": 986
},
{
"epoch": 1.1878386514148103,
"grad_norm": 0.6701137023889631,
"learning_rate": 3.8846227180701335e-05,
"loss": 0.6109538674354553,
"step": 987
},
{
"epoch": 1.1890427453341361,
"grad_norm": 0.5571428533929209,
"learning_rate": 3.874921423423697e-05,
"loss": 0.5311257839202881,
"step": 988
},
{
"epoch": 1.1902468392534618,
"grad_norm": 0.6933413663343239,
"learning_rate": 3.86522458809542e-05,
"loss": 0.6332836151123047,
"step": 989
},
{
"epoch": 1.1914509331727874,
"grad_norm": 0.5720889716251273,
"learning_rate": 3.855532250519307e-05,
"loss": 0.5061079263687134,
"step": 990
},
{
"epoch": 1.1926550270921132,
"grad_norm": 0.6337726108063414,
"learning_rate": 3.845844449111535e-05,
"loss": 0.6507538557052612,
"step": 991
},
{
"epoch": 1.1938591210114389,
"grad_norm": 0.6876564417388028,
"learning_rate": 3.8361612222703015e-05,
"loss": 0.603594183921814,
"step": 992
},
{
"epoch": 1.1950632149307645,
"grad_norm": 0.62693954157751,
"learning_rate": 3.826482608375671e-05,
"loss": 0.65156090259552,
"step": 993
},
{
"epoch": 1.1962673088500904,
"grad_norm": 0.6174813194954165,
"learning_rate": 3.816808645789425e-05,
"loss": 0.5988015532493591,
"step": 994
},
{
"epoch": 1.197471402769416,
"grad_norm": 0.6366707975431302,
"learning_rate": 3.8071393728549074e-05,
"loss": 0.6115190982818604,
"step": 995
},
{
"epoch": 1.1986754966887416,
"grad_norm": 0.5739976440086356,
"learning_rate": 3.79747482789688e-05,
"loss": 0.5073090195655823,
"step": 996
},
{
"epoch": 1.1998795906080675,
"grad_norm": 0.7299628970418212,
"learning_rate": 3.787815049221361e-05,
"loss": 0.6128576397895813,
"step": 997
},
{
"epoch": 1.2010836845273931,
"grad_norm": 0.5928975154363226,
"learning_rate": 3.778160075115476e-05,
"loss": 0.7040702104568481,
"step": 998
},
{
"epoch": 1.2022877784467187,
"grad_norm": 0.6764716165306945,
"learning_rate": 3.768509943847309e-05,
"loss": 0.6583726406097412,
"step": 999
},
{
"epoch": 1.2034918723660446,
"grad_norm": 0.7655399104406154,
"learning_rate": 3.758864693665748e-05,
"loss": 0.6048669815063477,
"step": 1000
},
{
"epoch": 1.2046959662853702,
"grad_norm": 0.6929632755591547,
"learning_rate": 3.749224362800338e-05,
"loss": 0.5911441445350647,
"step": 1001
},
{
"epoch": 1.205900060204696,
"grad_norm": 0.6678428093707727,
"learning_rate": 3.739588989461123e-05,
"loss": 0.5826472043991089,
"step": 1002
},
{
"epoch": 1.2071041541240217,
"grad_norm": 0.5742909832682589,
"learning_rate": 3.729958611838496e-05,
"loss": 0.5282671451568604,
"step": 1003
},
{
"epoch": 1.2083082480433474,
"grad_norm": 0.6593846672932838,
"learning_rate": 3.720333268103055e-05,
"loss": 0.6393536925315857,
"step": 1004
},
{
"epoch": 1.209512341962673,
"grad_norm": 0.645048734233651,
"learning_rate": 3.7107129964054366e-05,
"loss": 0.5161780118942261,
"step": 1005
},
{
"epoch": 1.2107164358819988,
"grad_norm": 0.6521270937616732,
"learning_rate": 3.701097834876185e-05,
"loss": 0.5652678608894348,
"step": 1006
},
{
"epoch": 1.2119205298013245,
"grad_norm": 0.719291801367232,
"learning_rate": 3.691487821625583e-05,
"loss": 0.6383004188537598,
"step": 1007
},
{
"epoch": 1.2131246237206503,
"grad_norm": 0.6723641088759535,
"learning_rate": 3.6818829947435076e-05,
"loss": 0.6614134311676025,
"step": 1008
},
{
"epoch": 1.214328717639976,
"grad_norm": 0.6446497075244803,
"learning_rate": 3.672283392299282e-05,
"loss": 0.5248739719390869,
"step": 1009
},
{
"epoch": 1.2155328115593016,
"grad_norm": 0.6777640011121145,
"learning_rate": 3.66268905234152e-05,
"loss": 0.6154916286468506,
"step": 1010
},
{
"epoch": 1.2167369054786272,
"grad_norm": 0.6609459633009753,
"learning_rate": 3.653100012897983e-05,
"loss": 0.6184529066085815,
"step": 1011
},
{
"epoch": 1.217940999397953,
"grad_norm": 0.6888385737563062,
"learning_rate": 3.643516311975413e-05,
"loss": 0.6395118236541748,
"step": 1012
},
{
"epoch": 1.2191450933172787,
"grad_norm": 0.6490466754073637,
"learning_rate": 3.633937987559402e-05,
"loss": 0.5476089715957642,
"step": 1013
},
{
"epoch": 1.2203491872366046,
"grad_norm": 0.7441213710432538,
"learning_rate": 3.624365077614226e-05,
"loss": 0.5452324748039246,
"step": 1014
},
{
"epoch": 1.2215532811559302,
"grad_norm": 0.6645629091016498,
"learning_rate": 3.614797620082703e-05,
"loss": 0.5184590220451355,
"step": 1015
},
{
"epoch": 1.2227573750752558,
"grad_norm": 0.7419621941545511,
"learning_rate": 3.605235652886042e-05,
"loss": 0.6348177194595337,
"step": 1016
},
{
"epoch": 1.2239614689945815,
"grad_norm": 0.6707127536069359,
"learning_rate": 3.595679213923685e-05,
"loss": 0.4938550591468811,
"step": 1017
},
{
"epoch": 1.2251655629139073,
"grad_norm": 0.6315829793918256,
"learning_rate": 3.586128341073167e-05,
"loss": 0.5749923586845398,
"step": 1018
},
{
"epoch": 1.226369656833233,
"grad_norm": 0.6331477125983097,
"learning_rate": 3.57658307218996e-05,
"loss": 0.5916865468025208,
"step": 1019
},
{
"epoch": 1.2275737507525588,
"grad_norm": 0.7846749860867426,
"learning_rate": 3.567043445107326e-05,
"loss": 0.5363922715187073,
"step": 1020
},
{
"epoch": 1.2287778446718844,
"grad_norm": 0.6965711425754788,
"learning_rate": 3.5575094976361625e-05,
"loss": 0.6076053380966187,
"step": 1021
},
{
"epoch": 1.22998193859121,
"grad_norm": 0.6685618608419817,
"learning_rate": 3.5479812675648575e-05,
"loss": 0.5980309844017029,
"step": 1022
},
{
"epoch": 1.2311860325105357,
"grad_norm": 0.5858609515515356,
"learning_rate": 3.5384587926591355e-05,
"loss": 0.5327468514442444,
"step": 1023
},
{
"epoch": 1.2323901264298616,
"grad_norm": 0.5918812910680888,
"learning_rate": 3.5289421106619126e-05,
"loss": 0.5000705718994141,
"step": 1024
},
{
"epoch": 1.2335942203491872,
"grad_norm": 0.6210027581503564,
"learning_rate": 3.519431259293147e-05,
"loss": 0.5952237248420715,
"step": 1025
},
{
"epoch": 1.234798314268513,
"grad_norm": 0.7034937909157433,
"learning_rate": 3.509926276249676e-05,
"loss": 0.6087906956672668,
"step": 1026
},
{
"epoch": 1.2360024081878387,
"grad_norm": 0.6839405989170235,
"learning_rate": 3.500427199205091e-05,
"loss": 0.5325309038162231,
"step": 1027
},
{
"epoch": 1.2372065021071643,
"grad_norm": 0.6182421717192286,
"learning_rate": 3.4909340658095646e-05,
"loss": 0.580743134021759,
"step": 1028
},
{
"epoch": 1.23841059602649,
"grad_norm": 0.7459897660008593,
"learning_rate": 3.4814469136897165e-05,
"loss": 0.6250452995300293,
"step": 1029
},
{
"epoch": 1.2396146899458158,
"grad_norm": 0.7099900777127174,
"learning_rate": 3.471965780448461e-05,
"loss": 0.6571633219718933,
"step": 1030
},
{
"epoch": 1.2408187838651414,
"grad_norm": 0.6553277230992378,
"learning_rate": 3.462490703664846e-05,
"loss": 0.49589502811431885,
"step": 1031
},
{
"epoch": 1.2420228777844673,
"grad_norm": 0.663723104863035,
"learning_rate": 3.4530217208939274e-05,
"loss": 0.5967923998832703,
"step": 1032
},
{
"epoch": 1.243226971703793,
"grad_norm": 0.709133317311018,
"learning_rate": 3.443558869666598e-05,
"loss": 0.6661368012428284,
"step": 1033
},
{
"epoch": 1.2444310656231186,
"grad_norm": 0.7061324636303485,
"learning_rate": 3.434102187489451e-05,
"loss": 0.6587533354759216,
"step": 1034
},
{
"epoch": 1.2456351595424442,
"grad_norm": 0.6579702542967999,
"learning_rate": 3.424651711844632e-05,
"loss": 0.5834370851516724,
"step": 1035
},
{
"epoch": 1.24683925346177,
"grad_norm": 0.6615924637591785,
"learning_rate": 3.415207480189676e-05,
"loss": 0.6136612892150879,
"step": 1036
},
{
"epoch": 1.2480433473810957,
"grad_norm": 0.6772710951274478,
"learning_rate": 3.405769529957381e-05,
"loss": 0.6056623458862305,
"step": 1037
},
{
"epoch": 1.2492474413004215,
"grad_norm": 0.642702675489295,
"learning_rate": 3.396337898555642e-05,
"loss": 0.6358840465545654,
"step": 1038
},
{
"epoch": 1.2504515352197472,
"grad_norm": 0.9252100643775776,
"learning_rate": 3.386912623367311e-05,
"loss": 0.5783637166023254,
"step": 1039
},
{
"epoch": 1.2516556291390728,
"grad_norm": 0.701983248076389,
"learning_rate": 3.377493741750044e-05,
"loss": 0.6720529198646545,
"step": 1040
},
{
"epoch": 1.2528597230583984,
"grad_norm": 0.6812048584013172,
"learning_rate": 3.36808129103616e-05,
"loss": 0.6228716373443604,
"step": 1041
},
{
"epoch": 1.2540638169777243,
"grad_norm": 0.6154647288628092,
"learning_rate": 3.358675308532486e-05,
"loss": 0.5664002895355225,
"step": 1042
},
{
"epoch": 1.25526791089705,
"grad_norm": 0.6241040220254938,
"learning_rate": 3.349275831520212e-05,
"loss": 0.5217874050140381,
"step": 1043
},
{
"epoch": 1.2564720048163758,
"grad_norm": 0.6562678876719462,
"learning_rate": 3.339882897254746e-05,
"loss": 0.5833320021629333,
"step": 1044
},
{
"epoch": 1.2576760987357014,
"grad_norm": 0.6095915930331086,
"learning_rate": 3.330496542965556e-05,
"loss": 0.6384044885635376,
"step": 1045
},
{
"epoch": 1.258880192655027,
"grad_norm": 0.6416377081327351,
"learning_rate": 3.3211168058560386e-05,
"loss": 0.5927673578262329,
"step": 1046
},
{
"epoch": 1.2600842865743527,
"grad_norm": 0.6100247656923646,
"learning_rate": 3.311743723103357e-05,
"loss": 0.6136247515678406,
"step": 1047
},
{
"epoch": 1.2612883804936785,
"grad_norm": 0.6317630200071538,
"learning_rate": 3.302377331858302e-05,
"loss": 0.6653052568435669,
"step": 1048
},
{
"epoch": 1.2624924744130042,
"grad_norm": 0.6274741674474991,
"learning_rate": 3.293017669245144e-05,
"loss": 0.5540253520011902,
"step": 1049
},
{
"epoch": 1.26369656833233,
"grad_norm": 0.6909319297738697,
"learning_rate": 3.283664772361476e-05,
"loss": 0.6327494382858276,
"step": 1050
},
{
"epoch": 1.2649006622516556,
"grad_norm": 0.6952322182958984,
"learning_rate": 3.274318678278085e-05,
"loss": 0.6329002380371094,
"step": 1051
},
{
"epoch": 1.2661047561709813,
"grad_norm": 0.6308869283921665,
"learning_rate": 3.264979424038789e-05,
"loss": 0.5057480335235596,
"step": 1052
},
{
"epoch": 1.267308850090307,
"grad_norm": 0.6285730530315006,
"learning_rate": 3.255647046660296e-05,
"loss": 0.6263934969902039,
"step": 1053
},
{
"epoch": 1.2685129440096328,
"grad_norm": 0.6400200018215665,
"learning_rate": 3.246321583132058e-05,
"loss": 0.6725963950157166,
"step": 1054
},
{
"epoch": 1.2697170379289584,
"grad_norm": 0.7099091081910462,
"learning_rate": 3.237003070416122e-05,
"loss": 0.6018358469009399,
"step": 1055
},
{
"epoch": 1.2709211318482843,
"grad_norm": 0.6135782078162696,
"learning_rate": 3.22769154544699e-05,
"loss": 0.5996497869491577,
"step": 1056
},
{
"epoch": 1.27212522576761,
"grad_norm": 0.6866426223054635,
"learning_rate": 3.2183870451314624e-05,
"loss": 0.5664709806442261,
"step": 1057
},
{
"epoch": 1.2733293196869355,
"grad_norm": 0.6090505212825614,
"learning_rate": 3.2090896063485e-05,
"loss": 0.5572994351387024,
"step": 1058
},
{
"epoch": 1.2745334136062612,
"grad_norm": 0.6372391798232508,
"learning_rate": 3.19979926594907e-05,
"loss": 0.5810008645057678,
"step": 1059
},
{
"epoch": 1.275737507525587,
"grad_norm": 0.648696717169655,
"learning_rate": 3.190516060756009e-05,
"loss": 0.5552839040756226,
"step": 1060
},
{
"epoch": 1.2769416014449126,
"grad_norm": 0.6031159136239559,
"learning_rate": 3.181240027563875e-05,
"loss": 0.5894042253494263,
"step": 1061
},
{
"epoch": 1.2781456953642385,
"grad_norm": 0.6774291943197286,
"learning_rate": 3.171971203138795e-05,
"loss": 0.678736686706543,
"step": 1062
},
{
"epoch": 1.2793497892835641,
"grad_norm": 0.609896680228515,
"learning_rate": 3.1627096242183243e-05,
"loss": 0.5225570201873779,
"step": 1063
},
{
"epoch": 1.2805538832028898,
"grad_norm": 0.679157586878992,
"learning_rate": 3.1534553275112994e-05,
"loss": 0.5883038640022278,
"step": 1064
},
{
"epoch": 1.2817579771222154,
"grad_norm": 0.6905787372952888,
"learning_rate": 3.144208349697696e-05,
"loss": 0.6585335731506348,
"step": 1065
},
{
"epoch": 1.2829620710415413,
"grad_norm": 0.6058564268171939,
"learning_rate": 3.13496872742848e-05,
"loss": 0.5555436611175537,
"step": 1066
},
{
"epoch": 1.2841661649608669,
"grad_norm": 0.6887579235137145,
"learning_rate": 3.125736497325464e-05,
"loss": 0.6298503279685974,
"step": 1067
},
{
"epoch": 1.2853702588801927,
"grad_norm": 0.6477445948330529,
"learning_rate": 3.116511695981157e-05,
"loss": 0.5965017676353455,
"step": 1068
},
{
"epoch": 1.2865743527995184,
"grad_norm": 0.6675640245384982,
"learning_rate": 3.107294359958628e-05,
"loss": 0.5505380034446716,
"step": 1069
},
{
"epoch": 1.287778446718844,
"grad_norm": 0.6437167909226186,
"learning_rate": 3.098084525791356e-05,
"loss": 0.6020991802215576,
"step": 1070
},
{
"epoch": 1.2889825406381699,
"grad_norm": 0.6633990777410962,
"learning_rate": 3.0888822299830854e-05,
"loss": 0.552418053150177,
"step": 1071
},
{
"epoch": 1.2901866345574955,
"grad_norm": 0.6046857201539841,
"learning_rate": 3.079687509007682e-05,
"loss": 0.5754314064979553,
"step": 1072
},
{
"epoch": 1.2913907284768211,
"grad_norm": 0.6971859400045242,
"learning_rate": 3.070500399308987e-05,
"loss": 0.6145979762077332,
"step": 1073
},
{
"epoch": 1.292594822396147,
"grad_norm": 0.6949536669306756,
"learning_rate": 3.0613209373006745e-05,
"loss": 0.6195294260978699,
"step": 1074
},
{
"epoch": 1.2937989163154726,
"grad_norm": 0.6242972237847882,
"learning_rate": 3.052149159366109e-05,
"loss": 0.5844000577926636,
"step": 1075
},
{
"epoch": 1.2950030102347982,
"grad_norm": 0.7035332273264987,
"learning_rate": 3.0429851018581955e-05,
"loss": 0.5531705617904663,
"step": 1076
},
{
"epoch": 1.296207104154124,
"grad_norm": 0.632115068867665,
"learning_rate": 3.0338288010992398e-05,
"loss": 0.5232405066490173,
"step": 1077
},
{
"epoch": 1.2974111980734497,
"grad_norm": 0.6125142431442002,
"learning_rate": 3.024680293380804e-05,
"loss": 0.53972327709198,
"step": 1078
},
{
"epoch": 1.2986152919927754,
"grad_norm": 0.7166615692901409,
"learning_rate": 3.0155396149635585e-05,
"loss": 0.5764614343643188,
"step": 1079
},
{
"epoch": 1.2998193859121012,
"grad_norm": 0.6590577657401502,
"learning_rate": 3.0064068020771486e-05,
"loss": 0.5957779884338379,
"step": 1080
},
{
"epoch": 1.3010234798314269,
"grad_norm": 0.6383819520827908,
"learning_rate": 2.9972818909200396e-05,
"loss": 0.5310415029525757,
"step": 1081
},
{
"epoch": 1.3022275737507525,
"grad_norm": 0.6264074933268168,
"learning_rate": 2.988164917659375e-05,
"loss": 0.5686416029930115,
"step": 1082
},
{
"epoch": 1.3034316676700783,
"grad_norm": 0.5735237535889868,
"learning_rate": 2.979055918430842e-05,
"loss": 0.6153852939605713,
"step": 1083
},
{
"epoch": 1.304635761589404,
"grad_norm": 0.6338428525876928,
"learning_rate": 2.9699549293385176e-05,
"loss": 0.5895254611968994,
"step": 1084
},
{
"epoch": 1.3058398555087296,
"grad_norm": 0.6628514943014948,
"learning_rate": 2.9608619864547337e-05,
"loss": 0.5714896321296692,
"step": 1085
},
{
"epoch": 1.3070439494280555,
"grad_norm": 0.6498256880211087,
"learning_rate": 2.95177712581993e-05,
"loss": 0.637000560760498,
"step": 1086
},
{
"epoch": 1.308248043347381,
"grad_norm": 0.5805140898940456,
"learning_rate": 2.9427003834425075e-05,
"loss": 0.5090338587760925,
"step": 1087
},
{
"epoch": 1.3094521372667067,
"grad_norm": 0.6568092367082321,
"learning_rate": 2.9336317952986946e-05,
"loss": 0.5969448089599609,
"step": 1088
},
{
"epoch": 1.3106562311860326,
"grad_norm": 0.6422165737966679,
"learning_rate": 2.924571397332398e-05,
"loss": 0.6276159286499023,
"step": 1089
},
{
"epoch": 1.3118603251053582,
"grad_norm": 0.6264592124277943,
"learning_rate": 2.915519225455065e-05,
"loss": 0.5930631756782532,
"step": 1090
},
{
"epoch": 1.3130644190246838,
"grad_norm": 0.760451843468021,
"learning_rate": 2.906475315545532e-05,
"loss": 0.5920497179031372,
"step": 1091
},
{
"epoch": 1.3142685129440097,
"grad_norm": 0.7376228210515698,
"learning_rate": 2.8974397034498917e-05,
"loss": 0.6203291416168213,
"step": 1092
},
{
"epoch": 1.3154726068633353,
"grad_norm": 0.6311093043141208,
"learning_rate": 2.8884124249813526e-05,
"loss": 0.5911815166473389,
"step": 1093
},
{
"epoch": 1.316676700782661,
"grad_norm": 0.6725958475028008,
"learning_rate": 2.879393515920087e-05,
"loss": 0.5948607325553894,
"step": 1094
},
{
"epoch": 1.3178807947019868,
"grad_norm": 0.6026740962582275,
"learning_rate": 2.8703830120130976e-05,
"loss": 0.5619141459465027,
"step": 1095
},
{
"epoch": 1.3190848886213125,
"grad_norm": 0.7327869849262203,
"learning_rate": 2.8613809489740662e-05,
"loss": 0.5664507150650024,
"step": 1096
},
{
"epoch": 1.320288982540638,
"grad_norm": 0.6133018339124032,
"learning_rate": 2.8523873624832247e-05,
"loss": 0.4463905692100525,
"step": 1097
},
{
"epoch": 1.321493076459964,
"grad_norm": 0.6412840762738142,
"learning_rate": 2.8434022881872103e-05,
"loss": 0.60805344581604,
"step": 1098
},
{
"epoch": 1.3226971703792896,
"grad_norm": 0.6492164235458658,
"learning_rate": 2.8344257616989144e-05,
"loss": 0.5742907524108887,
"step": 1099
},
{
"epoch": 1.3239012642986152,
"grad_norm": 0.634280196253906,
"learning_rate": 2.8254578185973556e-05,
"loss": 0.5517445206642151,
"step": 1100
},
{
"epoch": 1.325105358217941,
"grad_norm": 0.608061491520655,
"learning_rate": 2.8164984944275242e-05,
"loss": 0.5493748784065247,
"step": 1101
},
{
"epoch": 1.3263094521372667,
"grad_norm": 0.7070185171134912,
"learning_rate": 2.8075478247002518e-05,
"loss": 0.6650428175926208,
"step": 1102
},
{
"epoch": 1.3275135460565926,
"grad_norm": 0.7362968775469271,
"learning_rate": 2.7986058448920728e-05,
"loss": 0.6279030442237854,
"step": 1103
},
{
"epoch": 1.3287176399759182,
"grad_norm": 0.6587185876014878,
"learning_rate": 2.7896725904450748e-05,
"loss": 0.5652649402618408,
"step": 1104
},
{
"epoch": 1.3299217338952438,
"grad_norm": 0.6879422504214547,
"learning_rate": 2.7807480967667576e-05,
"loss": 0.5989348888397217,
"step": 1105
},
{
"epoch": 1.3311258278145695,
"grad_norm": 0.6023282107853354,
"learning_rate": 2.7718323992299e-05,
"loss": 0.5741807818412781,
"step": 1106
},
{
"epoch": 1.3323299217338953,
"grad_norm": 0.6239731759231483,
"learning_rate": 2.7629255331724225e-05,
"loss": 0.5270892977714539,
"step": 1107
},
{
"epoch": 1.333534015653221,
"grad_norm": 0.7508836444752833,
"learning_rate": 2.7540275338972343e-05,
"loss": 0.6074275374412537,
"step": 1108
},
{
"epoch": 1.3347381095725468,
"grad_norm": 0.6746453730710107,
"learning_rate": 2.7451384366721057e-05,
"loss": 0.6220452189445496,
"step": 1109
},
{
"epoch": 1.3359422034918724,
"grad_norm": 0.6766167757323907,
"learning_rate": 2.7362582767295158e-05,
"loss": 0.5820175409317017,
"step": 1110
},
{
"epoch": 1.337146297411198,
"grad_norm": 0.690797272196758,
"learning_rate": 2.7273870892665253e-05,
"loss": 0.5957927107810974,
"step": 1111
},
{
"epoch": 1.3383503913305237,
"grad_norm": 0.6430318040991272,
"learning_rate": 2.718524909444635e-05,
"loss": 0.5154853463172913,
"step": 1112
},
{
"epoch": 1.3395544852498495,
"grad_norm": 0.6582915506900874,
"learning_rate": 2.709671772389639e-05,
"loss": 0.5981411933898926,
"step": 1113
},
{
"epoch": 1.3407585791691752,
"grad_norm": 0.6590855324118988,
"learning_rate": 2.7008277131914916e-05,
"loss": 0.6747996211051941,
"step": 1114
},
{
"epoch": 1.341962673088501,
"grad_norm": 0.6448337036105954,
"learning_rate": 2.691992766904161e-05,
"loss": 0.5976516008377075,
"step": 1115
},
{
"epoch": 1.3431667670078267,
"grad_norm": 0.661465770553295,
"learning_rate": 2.683166968545503e-05,
"loss": 0.5724648833274841,
"step": 1116
},
{
"epoch": 1.3443708609271523,
"grad_norm": 0.7965142724805433,
"learning_rate": 2.6743503530971138e-05,
"loss": 0.6514641046524048,
"step": 1117
},
{
"epoch": 1.345574954846478,
"grad_norm": 0.6130309700978976,
"learning_rate": 2.6655429555041922e-05,
"loss": 0.524080753326416,
"step": 1118
},
{
"epoch": 1.3467790487658038,
"grad_norm": 0.676879651641497,
"learning_rate": 2.656744810675397e-05,
"loss": 0.5969399809837341,
"step": 1119
},
{
"epoch": 1.3479831426851294,
"grad_norm": 0.6443187750796282,
"learning_rate": 2.6479559534827168e-05,
"loss": 0.5908799171447754,
"step": 1120
},
{
"epoch": 1.3491872366044553,
"grad_norm": 0.7293081468185466,
"learning_rate": 2.6391764187613277e-05,
"loss": 0.639958381652832,
"step": 1121
},
{
"epoch": 1.350391330523781,
"grad_norm": 0.7245543963855401,
"learning_rate": 2.6304062413094588e-05,
"loss": 0.5507534742355347,
"step": 1122
},
{
"epoch": 1.3515954244431065,
"grad_norm": 0.7138419286324111,
"learning_rate": 2.6216454558882486e-05,
"loss": 0.5698534846305847,
"step": 1123
},
{
"epoch": 1.3527995183624322,
"grad_norm": 0.658594597820022,
"learning_rate": 2.6128940972216044e-05,
"loss": 0.6738436818122864,
"step": 1124
},
{
"epoch": 1.354003612281758,
"grad_norm": 0.6684797667576053,
"learning_rate": 2.604152199996077e-05,
"loss": 0.5904508829116821,
"step": 1125
},
{
"epoch": 1.3552077062010837,
"grad_norm": 0.6642004437577558,
"learning_rate": 2.5954197988607133e-05,
"loss": 0.6535474061965942,
"step": 1126
},
{
"epoch": 1.3564118001204095,
"grad_norm": 0.6007990769985496,
"learning_rate": 2.586696928426926e-05,
"loss": 0.6121514439582825,
"step": 1127
},
{
"epoch": 1.3576158940397351,
"grad_norm": 0.6336515572591528,
"learning_rate": 2.5779836232683442e-05,
"loss": 0.5832182765007019,
"step": 1128
},
{
"epoch": 1.3588199879590608,
"grad_norm": 0.6421286075260746,
"learning_rate": 2.5692799179206906e-05,
"loss": 0.5742759704589844,
"step": 1129
},
{
"epoch": 1.3600240818783864,
"grad_norm": 0.690150823325512,
"learning_rate": 2.5605858468816358e-05,
"loss": 0.580817461013794,
"step": 1130
},
{
"epoch": 1.3612281757977123,
"grad_norm": 0.6896471517754647,
"learning_rate": 2.5519014446106682e-05,
"loss": 0.588418185710907,
"step": 1131
},
{
"epoch": 1.362432269717038,
"grad_norm": 0.613216584761024,
"learning_rate": 2.5432267455289503e-05,
"loss": 0.5638326406478882,
"step": 1132
},
{
"epoch": 1.3636363636363638,
"grad_norm": 0.6210439632532312,
"learning_rate": 2.5345617840191828e-05,
"loss": 0.5994534492492676,
"step": 1133
},
{
"epoch": 1.3648404575556894,
"grad_norm": 0.6505033824932888,
"learning_rate": 2.5259065944254746e-05,
"loss": 0.6377952098846436,
"step": 1134
},
{
"epoch": 1.366044551475015,
"grad_norm": 0.6974309419675122,
"learning_rate": 2.5172612110532012e-05,
"loss": 0.6890112161636353,
"step": 1135
},
{
"epoch": 1.3672486453943407,
"grad_norm": 0.6202436590581667,
"learning_rate": 2.5086256681688745e-05,
"loss": 0.5471165180206299,
"step": 1136
},
{
"epoch": 1.3684527393136665,
"grad_norm": 0.5797278347327742,
"learning_rate": 2.500000000000001e-05,
"loss": 0.5749981999397278,
"step": 1137
},
{
"epoch": 1.3696568332329921,
"grad_norm": 0.6710778824322429,
"learning_rate": 2.491384240734943e-05,
"loss": 0.6004801988601685,
"step": 1138
},
{
"epoch": 1.370860927152318,
"grad_norm": 0.6858811733202708,
"learning_rate": 2.4827784245227937e-05,
"loss": 0.6247427463531494,
"step": 1139
},
{
"epoch": 1.3720650210716436,
"grad_norm": 0.6422633870461217,
"learning_rate": 2.4741825854732344e-05,
"loss": 0.574191689491272,
"step": 1140
},
{
"epoch": 1.3732691149909693,
"grad_norm": 0.6357294018183007,
"learning_rate": 2.4655967576564064e-05,
"loss": 0.5395843386650085,
"step": 1141
},
{
"epoch": 1.374473208910295,
"grad_norm": 0.6263451518446398,
"learning_rate": 2.4570209751027623e-05,
"loss": 0.5604453086853027,
"step": 1142
},
{
"epoch": 1.3756773028296208,
"grad_norm": 0.678607658650321,
"learning_rate": 2.4484552718029448e-05,
"loss": 0.6286793947219849,
"step": 1143
},
{
"epoch": 1.3768813967489464,
"grad_norm": 0.6734943023110295,
"learning_rate": 2.439899681707646e-05,
"loss": 0.6490384340286255,
"step": 1144
},
{
"epoch": 1.3780854906682722,
"grad_norm": 0.6232256600206579,
"learning_rate": 2.4313542387274717e-05,
"loss": 0.56013023853302,
"step": 1145
},
{
"epoch": 1.3792895845875979,
"grad_norm": 0.6512462577282983,
"learning_rate": 2.4228189767328173e-05,
"loss": 0.5262826085090637,
"step": 1146
},
{
"epoch": 1.3804936785069235,
"grad_norm": 0.6534793889464704,
"learning_rate": 2.4142939295537126e-05,
"loss": 0.5833800435066223,
"step": 1147
},
{
"epoch": 1.3816977724262491,
"grad_norm": 0.6376912481118475,
"learning_rate": 2.405779130979709e-05,
"loss": 0.5151797533035278,
"step": 1148
},
{
"epoch": 1.382901866345575,
"grad_norm": 0.6433663845535533,
"learning_rate": 2.3972746147597335e-05,
"loss": 0.5034987926483154,
"step": 1149
},
{
"epoch": 1.3841059602649006,
"grad_norm": 0.6387226460653036,
"learning_rate": 2.388780414601959e-05,
"loss": 0.6382157206535339,
"step": 1150
},
{
"epoch": 1.3853100541842265,
"grad_norm": 0.6525422712893322,
"learning_rate": 2.380296564173674e-05,
"loss": 0.5377070903778076,
"step": 1151
},
{
"epoch": 1.3865141481035521,
"grad_norm": 0.6361638026870527,
"learning_rate": 2.3718230971011386e-05,
"loss": 0.553045392036438,
"step": 1152
},
{
"epoch": 1.3877182420228777,
"grad_norm": 0.6604901502553285,
"learning_rate": 2.3633600469694606e-05,
"loss": 0.5241578817367554,
"step": 1153
},
{
"epoch": 1.3889223359422034,
"grad_norm": 0.6609889554265359,
"learning_rate": 2.3549074473224612e-05,
"loss": 0.5158299207687378,
"step": 1154
},
{
"epoch": 1.3901264298615292,
"grad_norm": 0.681790546788809,
"learning_rate": 2.3464653316625385e-05,
"loss": 0.5271784067153931,
"step": 1155
},
{
"epoch": 1.3913305237808549,
"grad_norm": 0.6429951344093262,
"learning_rate": 2.3380337334505374e-05,
"loss": 0.6050268411636353,
"step": 1156
},
{
"epoch": 1.3925346177001807,
"grad_norm": 0.6927991853136467,
"learning_rate": 2.3296126861056167e-05,
"loss": 0.5803988575935364,
"step": 1157
},
{
"epoch": 1.3937387116195064,
"grad_norm": 0.6355445702814094,
"learning_rate": 2.3212022230051154e-05,
"loss": 0.545645534992218,
"step": 1158
},
{
"epoch": 1.394942805538832,
"grad_norm": 0.6992209580511681,
"learning_rate": 2.3128023774844194e-05,
"loss": 0.6576249003410339,
"step": 1159
},
{
"epoch": 1.3961468994581576,
"grad_norm": 0.7005498664256508,
"learning_rate": 2.3044131828368386e-05,
"loss": 0.582336962223053,
"step": 1160
},
{
"epoch": 1.3973509933774835,
"grad_norm": 0.70377493838526,
"learning_rate": 2.2960346723134575e-05,
"loss": 0.5340085029602051,
"step": 1161
},
{
"epoch": 1.398555087296809,
"grad_norm": 0.6419337587745647,
"learning_rate": 2.2876668791230193e-05,
"loss": 0.6333436369895935,
"step": 1162
},
{
"epoch": 1.399759181216135,
"grad_norm": 0.7263202551262201,
"learning_rate": 2.2793098364317868e-05,
"loss": 0.6011248826980591,
"step": 1163
},
{
"epoch": 1.4009632751354606,
"grad_norm": 0.7532824340862169,
"learning_rate": 2.2709635773634113e-05,
"loss": 0.5780522227287292,
"step": 1164
},
{
"epoch": 1.4021673690547862,
"grad_norm": 0.6334080866576622,
"learning_rate": 2.2626281349988103e-05,
"loss": 0.62692791223526,
"step": 1165
},
{
"epoch": 1.4033714629741119,
"grad_norm": 0.6900123881398824,
"learning_rate": 2.254303542376016e-05,
"loss": 0.629398763179779,
"step": 1166
},
{
"epoch": 1.4045755568934377,
"grad_norm": 0.8266154500851933,
"learning_rate": 2.2459898324900662e-05,
"loss": 0.5797542929649353,
"step": 1167
},
{
"epoch": 1.4057796508127633,
"grad_norm": 0.6600289647001534,
"learning_rate": 2.2376870382928607e-05,
"loss": 0.5778429508209229,
"step": 1168
},
{
"epoch": 1.4069837447320892,
"grad_norm": 0.6559688369671797,
"learning_rate": 2.2293951926930356e-05,
"loss": 0.5707594156265259,
"step": 1169
},
{
"epoch": 1.4081878386514148,
"grad_norm": 0.6653841370747816,
"learning_rate": 2.2211143285558312e-05,
"loss": 0.6159984469413757,
"step": 1170
},
{
"epoch": 1.4093919325707405,
"grad_norm": 0.7649472065378574,
"learning_rate": 2.2128444787029618e-05,
"loss": 0.6222115755081177,
"step": 1171
},
{
"epoch": 1.410596026490066,
"grad_norm": 0.6796942807832598,
"learning_rate": 2.204585675912486e-05,
"loss": 0.6199472546577454,
"step": 1172
},
{
"epoch": 1.411800120409392,
"grad_norm": 0.6199548741185632,
"learning_rate": 2.1963379529186768e-05,
"loss": 0.590892493724823,
"step": 1173
},
{
"epoch": 1.4130042143287176,
"grad_norm": 0.6187634070466261,
"learning_rate": 2.1881013424118922e-05,
"loss": 0.5503128170967102,
"step": 1174
},
{
"epoch": 1.4142083082480434,
"grad_norm": 0.6512488781347366,
"learning_rate": 2.1798758770384442e-05,
"loss": 0.5549368858337402,
"step": 1175
},
{
"epoch": 1.415412402167369,
"grad_norm": 0.6347054155507705,
"learning_rate": 2.171661589400471e-05,
"loss": 0.6251223087310791,
"step": 1176
},
{
"epoch": 1.4166164960866947,
"grad_norm": 0.6695819615416071,
"learning_rate": 2.1634585120558078e-05,
"loss": 0.5890984535217285,
"step": 1177
},
{
"epoch": 1.4178205900060203,
"grad_norm": 0.6814695161224296,
"learning_rate": 2.1552666775178548e-05,
"loss": 0.5793978571891785,
"step": 1178
},
{
"epoch": 1.4190246839253462,
"grad_norm": 0.6726937714934542,
"learning_rate": 2.1470861182554534e-05,
"loss": 0.6175557971000671,
"step": 1179
},
{
"epoch": 1.4202287778446718,
"grad_norm": 0.6678641303443984,
"learning_rate": 2.138916866692754e-05,
"loss": 0.6240713596343994,
"step": 1180
},
{
"epoch": 1.4214328717639977,
"grad_norm": 0.6201792755242529,
"learning_rate": 2.1307589552090866e-05,
"loss": 0.5818585157394409,
"step": 1181
},
{
"epoch": 1.4226369656833233,
"grad_norm": 0.7015372658065818,
"learning_rate": 2.1226124161388354e-05,
"loss": 0.572142481803894,
"step": 1182
},
{
"epoch": 1.423841059602649,
"grad_norm": 0.6082394247396891,
"learning_rate": 2.1144772817713103e-05,
"loss": 0.4766699969768524,
"step": 1183
},
{
"epoch": 1.4250451535219746,
"grad_norm": 0.6387846065976666,
"learning_rate": 2.106353584350616e-05,
"loss": 0.6041740775108337,
"step": 1184
},
{
"epoch": 1.4262492474413004,
"grad_norm": 0.5982939961143208,
"learning_rate": 2.0982413560755272e-05,
"loss": 0.5946816205978394,
"step": 1185
},
{
"epoch": 1.427453341360626,
"grad_norm": 0.6688710695951081,
"learning_rate": 2.0901406290993598e-05,
"loss": 0.5802619457244873,
"step": 1186
},
{
"epoch": 1.428657435279952,
"grad_norm": 0.7078249109081977,
"learning_rate": 2.0820514355298432e-05,
"loss": 0.6126466989517212,
"step": 1187
},
{
"epoch": 1.4298615291992776,
"grad_norm": 0.6445823260166493,
"learning_rate": 2.073973807428995e-05,
"loss": 0.5297144651412964,
"step": 1188
},
{
"epoch": 1.4310656231186032,
"grad_norm": 0.6820450966327305,
"learning_rate": 2.0659077768129898e-05,
"loss": 0.6424897909164429,
"step": 1189
},
{
"epoch": 1.4322697170379288,
"grad_norm": 0.5940299387097965,
"learning_rate": 2.0578533756520375e-05,
"loss": 0.5096526741981506,
"step": 1190
},
{
"epoch": 1.4334738109572547,
"grad_norm": 0.5993494873406653,
"learning_rate": 2.049810635870251e-05,
"loss": 0.5202600955963135,
"step": 1191
},
{
"epoch": 1.4346779048765803,
"grad_norm": 0.7614707388132266,
"learning_rate": 2.0417795893455265e-05,
"loss": 0.6343950629234314,
"step": 1192
},
{
"epoch": 1.4358819987959062,
"grad_norm": 0.653826639908725,
"learning_rate": 2.0337602679094092e-05,
"loss": 0.605528712272644,
"step": 1193
},
{
"epoch": 1.4370860927152318,
"grad_norm": 0.7089308232671065,
"learning_rate": 2.0257527033469735e-05,
"loss": 0.6146092414855957,
"step": 1194
},
{
"epoch": 1.4382901866345574,
"grad_norm": 0.6770424284620841,
"learning_rate": 2.0177569273966945e-05,
"loss": 0.6413626670837402,
"step": 1195
},
{
"epoch": 1.439494280553883,
"grad_norm": 0.7544563351535553,
"learning_rate": 2.009772971750322e-05,
"loss": 0.5747498273849487,
"step": 1196
},
{
"epoch": 1.440698374473209,
"grad_norm": 0.6243897855893447,
"learning_rate": 2.0018008680527556e-05,
"loss": 0.5720393657684326,
"step": 1197
},
{
"epoch": 1.4419024683925346,
"grad_norm": 0.6904757741414711,
"learning_rate": 1.9938406479019183e-05,
"loss": 0.5748528838157654,
"step": 1198
},
{
"epoch": 1.4431065623118604,
"grad_norm": 0.6133136479702628,
"learning_rate": 1.985892342848633e-05,
"loss": 0.5648344159126282,
"step": 1199
},
{
"epoch": 1.444310656231186,
"grad_norm": 0.692341065686183,
"learning_rate": 1.9779559843964957e-05,
"loss": 0.5337613224983215,
"step": 1200
},
{
"epoch": 1.4455147501505117,
"grad_norm": 0.6932538183790291,
"learning_rate": 1.9700316040017515e-05,
"loss": 0.5030450224876404,
"step": 1201
},
{
"epoch": 1.4467188440698375,
"grad_norm": 0.6085144049389716,
"learning_rate": 1.9621192330731684e-05,
"loss": 0.5549513697624207,
"step": 1202
},
{
"epoch": 1.4479229379891632,
"grad_norm": 0.7071743976330627,
"learning_rate": 1.954218902971917e-05,
"loss": 0.6263749599456787,
"step": 1203
},
{
"epoch": 1.4491270319084888,
"grad_norm": 0.6532630278934997,
"learning_rate": 1.9463306450114416e-05,
"loss": 0.5744891166687012,
"step": 1204
},
{
"epoch": 1.4503311258278146,
"grad_norm": 0.6910032058096278,
"learning_rate": 1.938454490457337e-05,
"loss": 0.6153848767280579,
"step": 1205
},
{
"epoch": 1.4515352197471403,
"grad_norm": 0.6854934057332991,
"learning_rate": 1.9305904705272275e-05,
"loss": 0.6504592299461365,
"step": 1206
},
{
"epoch": 1.452739313666466,
"grad_norm": 0.6844756673065995,
"learning_rate": 1.922738616390639e-05,
"loss": 0.6322448253631592,
"step": 1207
},
{
"epoch": 1.4539434075857918,
"grad_norm": 0.5726761208441021,
"learning_rate": 1.9148989591688786e-05,
"loss": 0.5301163792610168,
"step": 1208
},
{
"epoch": 1.4551475015051174,
"grad_norm": 0.6499098121575463,
"learning_rate": 1.9070715299349105e-05,
"loss": 0.5792317390441895,
"step": 1209
},
{
"epoch": 1.456351595424443,
"grad_norm": 0.6294029092092295,
"learning_rate": 1.8992563597132323e-05,
"loss": 0.5840668082237244,
"step": 1210
},
{
"epoch": 1.457555689343769,
"grad_norm": 0.6288063174886138,
"learning_rate": 1.8914534794797513e-05,
"loss": 0.5666556358337402,
"step": 1211
},
{
"epoch": 1.4587597832630945,
"grad_norm": 0.6811900356717014,
"learning_rate": 1.8836629201616635e-05,
"loss": 0.6121422648429871,
"step": 1212
},
{
"epoch": 1.4599638771824202,
"grad_norm": 0.7039207316933149,
"learning_rate": 1.8758847126373303e-05,
"loss": 0.5998222827911377,
"step": 1213
},
{
"epoch": 1.461167971101746,
"grad_norm": 0.6245939644307974,
"learning_rate": 1.8681188877361555e-05,
"loss": 0.5631198883056641,
"step": 1214
},
{
"epoch": 1.4623720650210716,
"grad_norm": 0.6657764395909689,
"learning_rate": 1.8603654762384642e-05,
"loss": 0.5070215463638306,
"step": 1215
},
{
"epoch": 1.4635761589403973,
"grad_norm": 0.6946821767356586,
"learning_rate": 1.85262450887538e-05,
"loss": 0.5660195350646973,
"step": 1216
},
{
"epoch": 1.4647802528597231,
"grad_norm": 0.6486254208618857,
"learning_rate": 1.8448960163287034e-05,
"loss": 0.6307017207145691,
"step": 1217
},
{
"epoch": 1.4659843467790488,
"grad_norm": 0.651794455656358,
"learning_rate": 1.83718002923079e-05,
"loss": 0.6030329465866089,
"step": 1218
},
{
"epoch": 1.4671884406983744,
"grad_norm": 0.7355979968125673,
"learning_rate": 1.8294765781644285e-05,
"loss": 0.6677907109260559,
"step": 1219
},
{
"epoch": 1.4683925346177003,
"grad_norm": 0.6991027780768566,
"learning_rate": 1.821785693662724e-05,
"loss": 0.5083788633346558,
"step": 1220
},
{
"epoch": 1.4695966285370259,
"grad_norm": 0.6647763000946036,
"learning_rate": 1.8141074062089657e-05,
"loss": 0.603174090385437,
"step": 1221
},
{
"epoch": 1.4708007224563515,
"grad_norm": 0.7002973158395647,
"learning_rate": 1.8064417462365226e-05,
"loss": 0.5698137283325195,
"step": 1222
},
{
"epoch": 1.4720048163756774,
"grad_norm": 0.5833475144603689,
"learning_rate": 1.798788744128707e-05,
"loss": 0.5906316637992859,
"step": 1223
},
{
"epoch": 1.473208910295003,
"grad_norm": 0.6447305050041968,
"learning_rate": 1.791148430218666e-05,
"loss": 0.6068233847618103,
"step": 1224
},
{
"epoch": 1.4744130042143286,
"grad_norm": 0.6277096478361395,
"learning_rate": 1.7835208347892535e-05,
"loss": 0.5847250819206238,
"step": 1225
},
{
"epoch": 1.4756170981336545,
"grad_norm": 0.7414175683459697,
"learning_rate": 1.775905988072909e-05,
"loss": 0.5115228295326233,
"step": 1226
},
{
"epoch": 1.4768211920529801,
"grad_norm": 0.697043493410439,
"learning_rate": 1.768303920251551e-05,
"loss": 0.5767746567726135,
"step": 1227
},
{
"epoch": 1.4780252859723058,
"grad_norm": 0.6337907702044968,
"learning_rate": 1.7607146614564418e-05,
"loss": 0.5868975520133972,
"step": 1228
},
{
"epoch": 1.4792293798916316,
"grad_norm": 0.6840890488968902,
"learning_rate": 1.753138241768075e-05,
"loss": 0.5457339882850647,
"step": 1229
},
{
"epoch": 1.4804334738109572,
"grad_norm": 0.6499123239023711,
"learning_rate": 1.7455746912160564e-05,
"loss": 0.551403284072876,
"step": 1230
},
{
"epoch": 1.4816375677302829,
"grad_norm": 0.6101386305043538,
"learning_rate": 1.7380240397789836e-05,
"loss": 0.5619639158248901,
"step": 1231
},
{
"epoch": 1.4828416616496087,
"grad_norm": 0.7183606920392251,
"learning_rate": 1.7304863173843283e-05,
"loss": 0.5660442113876343,
"step": 1232
},
{
"epoch": 1.4840457555689344,
"grad_norm": 0.6850065823958903,
"learning_rate": 1.722961553908318e-05,
"loss": 0.6238867044448853,
"step": 1233
},
{
"epoch": 1.4852498494882602,
"grad_norm": 0.716310161029593,
"learning_rate": 1.7154497791758157e-05,
"loss": 0.6281149387359619,
"step": 1234
},
{
"epoch": 1.4864539434075859,
"grad_norm": 0.6551910228833004,
"learning_rate": 1.707951022960199e-05,
"loss": 0.6366140246391296,
"step": 1235
},
{
"epoch": 1.4876580373269115,
"grad_norm": 0.6875009342751863,
"learning_rate": 1.700465314983254e-05,
"loss": 0.5912625789642334,
"step": 1236
},
{
"epoch": 1.4888621312462371,
"grad_norm": 0.7103107098011074,
"learning_rate": 1.6929926849150428e-05,
"loss": 0.507576584815979,
"step": 1237
},
{
"epoch": 1.490066225165563,
"grad_norm": 0.7785689863841994,
"learning_rate": 1.6855331623737958e-05,
"loss": 0.5702663064002991,
"step": 1238
},
{
"epoch": 1.4912703190848886,
"grad_norm": 0.7085393061941329,
"learning_rate": 1.6780867769257898e-05,
"loss": 0.6586170792579651,
"step": 1239
},
{
"epoch": 1.4924744130042145,
"grad_norm": 0.6929869277646624,
"learning_rate": 1.6706535580852267e-05,
"loss": 0.6001737117767334,
"step": 1240
},
{
"epoch": 1.49367850692354,
"grad_norm": 0.619473871811379,
"learning_rate": 1.6632335353141333e-05,
"loss": 0.5791480541229248,
"step": 1241
},
{
"epoch": 1.4948826008428657,
"grad_norm": 0.7486550213430624,
"learning_rate": 1.6558267380222224e-05,
"loss": 0.6321637630462646,
"step": 1242
},
{
"epoch": 1.4960866947621914,
"grad_norm": 0.6112954966976807,
"learning_rate": 1.6484331955667947e-05,
"loss": 0.5635457634925842,
"step": 1243
},
{
"epoch": 1.4972907886815172,
"grad_norm": 0.7108878803102963,
"learning_rate": 1.6410529372526057e-05,
"loss": 0.7190176248550415,
"step": 1244
},
{
"epoch": 1.4984948826008428,
"grad_norm": 0.6658849135083088,
"learning_rate": 1.6336859923317643e-05,
"loss": 0.5681280493736267,
"step": 1245
},
{
"epoch": 1.4996989765201687,
"grad_norm": 0.6118879060848942,
"learning_rate": 1.6263323900036126e-05,
"loss": 0.5625150203704834,
"step": 1246
},
{
"epoch": 1.5009030704394943,
"grad_norm": 0.643240167423122,
"learning_rate": 1.6189921594146046e-05,
"loss": 0.5696181058883667,
"step": 1247
},
{
"epoch": 1.50210716435882,
"grad_norm": 0.5934555486357125,
"learning_rate": 1.6116653296581975e-05,
"loss": 0.5061366558074951,
"step": 1248
},
{
"epoch": 1.5033112582781456,
"grad_norm": 0.6301639218719404,
"learning_rate": 1.6043519297747285e-05,
"loss": 0.5271162986755371,
"step": 1249
},
{
"epoch": 1.5045153521974715,
"grad_norm": 0.6558350418754424,
"learning_rate": 1.5970519887513073e-05,
"loss": 0.5793251991271973,
"step": 1250
},
{
"epoch": 1.505719446116797,
"grad_norm": 0.6166848173267853,
"learning_rate": 1.5897655355217038e-05,
"loss": 0.49026674032211304,
"step": 1251
},
{
"epoch": 1.506923540036123,
"grad_norm": 0.639230504451563,
"learning_rate": 1.5824925989662216e-05,
"loss": 0.4468444585800171,
"step": 1252
},
{
"epoch": 1.5081276339554486,
"grad_norm": 0.6279029266897213,
"learning_rate": 1.5752332079115932e-05,
"loss": 0.5121999979019165,
"step": 1253
},
{
"epoch": 1.5093317278747742,
"grad_norm": 0.6897589017387046,
"learning_rate": 1.5679873911308597e-05,
"loss": 0.6195895075798035,
"step": 1254
},
{
"epoch": 1.5105358217940998,
"grad_norm": 0.6405551867527671,
"learning_rate": 1.56075517734326e-05,
"loss": 0.6228514909744263,
"step": 1255
},
{
"epoch": 1.5117399157134257,
"grad_norm": 0.6712649126230317,
"learning_rate": 1.5535365952141233e-05,
"loss": 0.626694917678833,
"step": 1256
},
{
"epoch": 1.5129440096327513,
"grad_norm": 0.6944075950230268,
"learning_rate": 1.5463316733547434e-05,
"loss": 0.6293349266052246,
"step": 1257
},
{
"epoch": 1.5141481035520772,
"grad_norm": 0.6474800499836418,
"learning_rate": 1.5391404403222676e-05,
"loss": 0.4614093601703644,
"step": 1258
},
{
"epoch": 1.5153521974714028,
"grad_norm": 0.6354596486501298,
"learning_rate": 1.5319629246195917e-05,
"loss": 0.5340675115585327,
"step": 1259
},
{
"epoch": 1.5165562913907285,
"grad_norm": 0.76231174162437,
"learning_rate": 1.5247991546952428e-05,
"loss": 0.5091834664344788,
"step": 1260
},
{
"epoch": 1.517760385310054,
"grad_norm": 0.7027773645202476,
"learning_rate": 1.5176491589432628e-05,
"loss": 0.5760695338249207,
"step": 1261
},
{
"epoch": 1.51896447922938,
"grad_norm": 0.6909468225730208,
"learning_rate": 1.5105129657031009e-05,
"loss": 0.5917767882347107,
"step": 1262
},
{
"epoch": 1.5201685731487056,
"grad_norm": 0.6413099265304232,
"learning_rate": 1.5033906032594958e-05,
"loss": 0.48429784178733826,
"step": 1263
},
{
"epoch": 1.5213726670680314,
"grad_norm": 0.7402093135988407,
"learning_rate": 1.4962820998423683e-05,
"loss": 0.5737481117248535,
"step": 1264
},
{
"epoch": 1.522576760987357,
"grad_norm": 0.6633749969252996,
"learning_rate": 1.4891874836267127e-05,
"loss": 0.5403265357017517,
"step": 1265
},
{
"epoch": 1.5237808549066827,
"grad_norm": 0.6703832368863357,
"learning_rate": 1.4821067827324753e-05,
"loss": 0.6162956953048706,
"step": 1266
},
{
"epoch": 1.5249849488260083,
"grad_norm": 0.6913767161943216,
"learning_rate": 1.4750400252244511e-05,
"loss": 0.5615054965019226,
"step": 1267
},
{
"epoch": 1.5261890427453342,
"grad_norm": 0.6431708429026083,
"learning_rate": 1.4679872391121651e-05,
"loss": 0.6342799663543701,
"step": 1268
},
{
"epoch": 1.5273931366646598,
"grad_norm": 0.6246696358486011,
"learning_rate": 1.4609484523497696e-05,
"loss": 0.4823196232318878,
"step": 1269
},
{
"epoch": 1.5285972305839857,
"grad_norm": 0.6643995241856853,
"learning_rate": 1.4539236928359318e-05,
"loss": 0.5989305377006531,
"step": 1270
},
{
"epoch": 1.5298013245033113,
"grad_norm": 0.6970680691115482,
"learning_rate": 1.4469129884137177e-05,
"loss": 0.5505509972572327,
"step": 1271
},
{
"epoch": 1.531005418422637,
"grad_norm": 0.7209244965404736,
"learning_rate": 1.4399163668704829e-05,
"loss": 0.5916901230812073,
"step": 1272
},
{
"epoch": 1.5322095123419626,
"grad_norm": 0.6672238207490072,
"learning_rate": 1.4329338559377691e-05,
"loss": 0.5727331042289734,
"step": 1273
},
{
"epoch": 1.5334136062612884,
"grad_norm": 0.7282879333042002,
"learning_rate": 1.4259654832911867e-05,
"loss": 0.6152413487434387,
"step": 1274
},
{
"epoch": 1.534617700180614,
"grad_norm": 0.6848054247552551,
"learning_rate": 1.4190112765503117e-05,
"loss": 0.6155499815940857,
"step": 1275
},
{
"epoch": 1.53582179409994,
"grad_norm": 0.6779262829361238,
"learning_rate": 1.412071263278571e-05,
"loss": 0.6265017986297607,
"step": 1276
},
{
"epoch": 1.5370258880192655,
"grad_norm": 0.6345807833091854,
"learning_rate": 1.4051454709831308e-05,
"loss": 0.5726587772369385,
"step": 1277
},
{
"epoch": 1.5382299819385912,
"grad_norm": 0.7476588469408837,
"learning_rate": 1.398233927114797e-05,
"loss": 0.6458104252815247,
"step": 1278
},
{
"epoch": 1.5394340758579168,
"grad_norm": 1.1233935398982429,
"learning_rate": 1.3913366590678966e-05,
"loss": 0.47828546166419983,
"step": 1279
},
{
"epoch": 1.5406381697772427,
"grad_norm": 0.6735195132494955,
"learning_rate": 1.38445369418018e-05,
"loss": 0.6620016694068909,
"step": 1280
},
{
"epoch": 1.5418422636965683,
"grad_norm": 0.7434175109937722,
"learning_rate": 1.377585059732701e-05,
"loss": 0.5846740007400513,
"step": 1281
},
{
"epoch": 1.5430463576158941,
"grad_norm": 0.6450732267312386,
"learning_rate": 1.370730782949713e-05,
"loss": 0.547637403011322,
"step": 1282
},
{
"epoch": 1.5442504515352198,
"grad_norm": 0.6742849341522323,
"learning_rate": 1.3638908909985621e-05,
"loss": 0.5787967443466187,
"step": 1283
},
{
"epoch": 1.5454545454545454,
"grad_norm": 0.63788147336325,
"learning_rate": 1.3570654109895853e-05,
"loss": 0.5432387590408325,
"step": 1284
},
{
"epoch": 1.546658639373871,
"grad_norm": 0.6879935367171848,
"learning_rate": 1.3502543699759917e-05,
"loss": 0.6357567310333252,
"step": 1285
},
{
"epoch": 1.547862733293197,
"grad_norm": 0.6466404497884245,
"learning_rate": 1.3434577949537592e-05,
"loss": 0.5758388638496399,
"step": 1286
},
{
"epoch": 1.5490668272125225,
"grad_norm": 0.7131752950787134,
"learning_rate": 1.336675712861532e-05,
"loss": 0.5336190462112427,
"step": 1287
},
{
"epoch": 1.5502709211318484,
"grad_norm": 0.6915472030489775,
"learning_rate": 1.3299081505805088e-05,
"loss": 0.5518936514854431,
"step": 1288
},
{
"epoch": 1.551475015051174,
"grad_norm": 0.7214042959852351,
"learning_rate": 1.3231551349343413e-05,
"loss": 0.6907586455345154,
"step": 1289
},
{
"epoch": 1.5526791089704997,
"grad_norm": 0.7321326805875424,
"learning_rate": 1.3164166926890226e-05,
"loss": 0.6275293827056885,
"step": 1290
},
{
"epoch": 1.5538832028898253,
"grad_norm": 0.6243437937781886,
"learning_rate": 1.3096928505527811e-05,
"loss": 0.6157265901565552,
"step": 1291
},
{
"epoch": 1.5550872968091511,
"grad_norm": 0.7585166939999731,
"learning_rate": 1.3029836351759795e-05,
"loss": 0.6015262007713318,
"step": 1292
},
{
"epoch": 1.5562913907284768,
"grad_norm": 0.6667612051008648,
"learning_rate": 1.296289073151004e-05,
"loss": 0.5566407442092896,
"step": 1293
},
{
"epoch": 1.5574954846478026,
"grad_norm": 0.6470430817221635,
"learning_rate": 1.2896091910121666e-05,
"loss": 0.5622594356536865,
"step": 1294
},
{
"epoch": 1.5586995785671283,
"grad_norm": 0.6804382719854897,
"learning_rate": 1.2829440152355915e-05,
"loss": 0.5803040266036987,
"step": 1295
},
{
"epoch": 1.559903672486454,
"grad_norm": 0.6687207305578332,
"learning_rate": 1.2762935722391089e-05,
"loss": 0.59840327501297,
"step": 1296
},
{
"epoch": 1.5611077664057795,
"grad_norm": 0.6603082346009755,
"learning_rate": 1.2696578883821614e-05,
"loss": 0.5307196974754333,
"step": 1297
},
{
"epoch": 1.5623118603251054,
"grad_norm": 0.7359481197118585,
"learning_rate": 1.263036989965688e-05,
"loss": 0.6292601823806763,
"step": 1298
},
{
"epoch": 1.563515954244431,
"grad_norm": 0.630384514242762,
"learning_rate": 1.2564309032320315e-05,
"loss": 0.5643035769462585,
"step": 1299
},
{
"epoch": 1.5647200481637569,
"grad_norm": 0.6173040021348626,
"learning_rate": 1.2498396543648195e-05,
"loss": 0.5473842620849609,
"step": 1300
},
{
"epoch": 1.5659241420830825,
"grad_norm": 0.6733910286056267,
"learning_rate": 1.2432632694888724e-05,
"loss": 0.5461156964302063,
"step": 1301
},
{
"epoch": 1.5671282360024081,
"grad_norm": 0.6894664872560027,
"learning_rate": 1.2367017746700977e-05,
"loss": 0.551993191242218,
"step": 1302
},
{
"epoch": 1.5683323299217338,
"grad_norm": 0.6222609535272838,
"learning_rate": 1.2301551959153813e-05,
"loss": 0.4720235764980316,
"step": 1303
},
{
"epoch": 1.5695364238410596,
"grad_norm": 0.6885496364305319,
"learning_rate": 1.223623559172496e-05,
"loss": 0.6039336919784546,
"step": 1304
},
{
"epoch": 1.5707405177603853,
"grad_norm": 0.6287657401109025,
"learning_rate": 1.217106890329981e-05,
"loss": 0.5178000926971436,
"step": 1305
},
{
"epoch": 1.5719446116797111,
"grad_norm": 0.7517299077792008,
"learning_rate": 1.2106052152170561e-05,
"loss": 0.6386547684669495,
"step": 1306
},
{
"epoch": 1.5731487055990367,
"grad_norm": 0.6610070710050879,
"learning_rate": 1.204118559603511e-05,
"loss": 0.5045919418334961,
"step": 1307
},
{
"epoch": 1.5743527995183624,
"grad_norm": 0.6881944227439661,
"learning_rate": 1.1976469491996028e-05,
"loss": 0.5601974725723267,
"step": 1308
},
{
"epoch": 1.575556893437688,
"grad_norm": 0.6554510558095886,
"learning_rate": 1.1911904096559589e-05,
"loss": 0.5212211012840271,
"step": 1309
},
{
"epoch": 1.5767609873570139,
"grad_norm": 0.637358447537357,
"learning_rate": 1.1847489665634709e-05,
"loss": 0.5502785444259644,
"step": 1310
},
{
"epoch": 1.5779650812763397,
"grad_norm": 0.620386240593587,
"learning_rate": 1.1783226454531942e-05,
"loss": 0.5157536864280701,
"step": 1311
},
{
"epoch": 1.5791691751956654,
"grad_norm": 0.6585676339813008,
"learning_rate": 1.1719114717962476e-05,
"loss": 0.5165415406227112,
"step": 1312
},
{
"epoch": 1.580373269114991,
"grad_norm": 0.7493791152879317,
"learning_rate": 1.1655154710037153e-05,
"loss": 0.6232336759567261,
"step": 1313
},
{
"epoch": 1.5815773630343166,
"grad_norm": 0.6952382170280249,
"learning_rate": 1.1591346684265359e-05,
"loss": 0.6074533462524414,
"step": 1314
},
{
"epoch": 1.5827814569536423,
"grad_norm": 0.6890649127414126,
"learning_rate": 1.1527690893554156e-05,
"loss": 0.54258131980896,
"step": 1315
},
{
"epoch": 1.583985550872968,
"grad_norm": 0.7723371841016708,
"learning_rate": 1.1464187590207171e-05,
"loss": 0.5698084235191345,
"step": 1316
},
{
"epoch": 1.585189644792294,
"grad_norm": 0.6803846972242944,
"learning_rate": 1.1400837025923649e-05,
"loss": 0.5348570346832275,
"step": 1317
},
{
"epoch": 1.5863937387116196,
"grad_norm": 0.7012133606971119,
"learning_rate": 1.1337639451797494e-05,
"loss": 0.5426389575004578,
"step": 1318
},
{
"epoch": 1.5875978326309452,
"grad_norm": 0.6744539282883935,
"learning_rate": 1.1274595118316134e-05,
"loss": 0.5523807406425476,
"step": 1319
},
{
"epoch": 1.5888019265502709,
"grad_norm": 0.6537647651178288,
"learning_rate": 1.1211704275359675e-05,
"loss": 0.6136876344680786,
"step": 1320
},
{
"epoch": 1.5900060204695965,
"grad_norm": 0.7246448682395509,
"learning_rate": 1.1148967172199848e-05,
"loss": 0.57940274477005,
"step": 1321
},
{
"epoch": 1.5912101143889223,
"grad_norm": 0.7262129784220396,
"learning_rate": 1.1086384057499011e-05,
"loss": 0.6009513139724731,
"step": 1322
},
{
"epoch": 1.5924142083082482,
"grad_norm": 0.6920385306362808,
"learning_rate": 1.1023955179309203e-05,
"loss": 0.630897045135498,
"step": 1323
},
{
"epoch": 1.5936183022275738,
"grad_norm": 0.7137604613022566,
"learning_rate": 1.0961680785071116e-05,
"loss": 0.5550639629364014,
"step": 1324
},
{
"epoch": 1.5948223961468995,
"grad_norm": 0.7085346864243411,
"learning_rate": 1.0899561121613145e-05,
"loss": 0.5286394357681274,
"step": 1325
},
{
"epoch": 1.596026490066225,
"grad_norm": 0.8302518321935719,
"learning_rate": 1.0837596435150398e-05,
"loss": 0.5946967601776123,
"step": 1326
},
{
"epoch": 1.5972305839855507,
"grad_norm": 0.6668203019912043,
"learning_rate": 1.0775786971283725e-05,
"loss": 0.5410081148147583,
"step": 1327
},
{
"epoch": 1.5984346779048766,
"grad_norm": 0.7107307013054678,
"learning_rate": 1.0714132974998731e-05,
"loss": 0.6099376678466797,
"step": 1328
},
{
"epoch": 1.5996387718242024,
"grad_norm": 0.692321863693783,
"learning_rate": 1.065263469066483e-05,
"loss": 0.5954639315605164,
"step": 1329
},
{
"epoch": 1.600842865743528,
"grad_norm": 0.670559078654551,
"learning_rate": 1.0591292362034255e-05,
"loss": 0.6572100520133972,
"step": 1330
},
{
"epoch": 1.6020469596628537,
"grad_norm": 0.7359764923754903,
"learning_rate": 1.0530106232241099e-05,
"loss": 0.528759777545929,
"step": 1331
},
{
"epoch": 1.6032510535821793,
"grad_norm": 0.7315406198530844,
"learning_rate": 1.0469076543800355e-05,
"loss": 0.6525534391403198,
"step": 1332
},
{
"epoch": 1.604455147501505,
"grad_norm": 0.6550323597585707,
"learning_rate": 1.0408203538606948e-05,
"loss": 0.5543904304504395,
"step": 1333
},
{
"epoch": 1.6056592414208308,
"grad_norm": 0.7877187830505626,
"learning_rate": 1.0347487457934768e-05,
"loss": 0.6900970935821533,
"step": 1334
},
{
"epoch": 1.6068633353401567,
"grad_norm": 0.7022130623932448,
"learning_rate": 1.0286928542435754e-05,
"loss": 0.5940430164337158,
"step": 1335
},
{
"epoch": 1.6080674292594823,
"grad_norm": 0.7548095541194471,
"learning_rate": 1.0226527032138878e-05,
"loss": 0.6002829074859619,
"step": 1336
},
{
"epoch": 1.609271523178808,
"grad_norm": 0.675148279495715,
"learning_rate": 1.0166283166449253e-05,
"loss": 0.6131722331047058,
"step": 1337
},
{
"epoch": 1.6104756170981336,
"grad_norm": 0.7157880014315426,
"learning_rate": 1.0106197184147143e-05,
"loss": 0.4769801199436188,
"step": 1338
},
{
"epoch": 1.6116797110174592,
"grad_norm": 0.7070675214972796,
"learning_rate": 1.0046269323387036e-05,
"loss": 0.5769504308700562,
"step": 1339
},
{
"epoch": 1.612883804936785,
"grad_norm": 0.676711262352062,
"learning_rate": 9.986499821696693e-06,
"loss": 0.5734344720840454,
"step": 1340
},
{
"epoch": 1.614087898856111,
"grad_norm": 0.6195920357404767,
"learning_rate": 9.926888915976219e-06,
"loss": 0.505179762840271,
"step": 1341
},
{
"epoch": 1.6152919927754366,
"grad_norm": 0.639936444761267,
"learning_rate": 9.867436842497103e-06,
"loss": 0.45891082286834717,
"step": 1342
},
{
"epoch": 1.6164960866947622,
"grad_norm": 0.6857643487164867,
"learning_rate": 9.808143836901284e-06,
"loss": 0.6181744933128357,
"step": 1343
},
{
"epoch": 1.6177001806140878,
"grad_norm": 0.6358596598261053,
"learning_rate": 9.74901013420026e-06,
"loss": 0.5622484087944031,
"step": 1344
},
{
"epoch": 1.6189042745334135,
"grad_norm": 0.6306024458716856,
"learning_rate": 9.69003596877408e-06,
"loss": 0.5141526460647583,
"step": 1345
},
{
"epoch": 1.6201083684527393,
"grad_norm": 0.7029423890190829,
"learning_rate": 9.631221574370491e-06,
"loss": 0.6316851377487183,
"step": 1346
},
{
"epoch": 1.6213124623720652,
"grad_norm": 0.6940021087538228,
"learning_rate": 9.572567184103958e-06,
"loss": 0.6254806518554688,
"step": 1347
},
{
"epoch": 1.6225165562913908,
"grad_norm": 0.6404210309720665,
"learning_rate": 9.514073030454762e-06,
"loss": 0.5518223643302917,
"step": 1348
},
{
"epoch": 1.6237206502107164,
"grad_norm": 0.6804784668574658,
"learning_rate": 9.455739345268088e-06,
"loss": 0.6273688673973083,
"step": 1349
},
{
"epoch": 1.624924744130042,
"grad_norm": 0.6862326525850346,
"learning_rate": 9.39756635975308e-06,
"loss": 0.545397937297821,
"step": 1350
},
{
"epoch": 1.6261288380493677,
"grad_norm": 0.6998552571990925,
"learning_rate": 9.339554304481951e-06,
"loss": 0.6098823547363281,
"step": 1351
},
{
"epoch": 1.6273329319686936,
"grad_norm": 0.6602839846160563,
"learning_rate": 9.281703409389043e-06,
"loss": 0.5640820264816284,
"step": 1352
},
{
"epoch": 1.6285370258880194,
"grad_norm": 0.6404985903308311,
"learning_rate": 9.224013903769946e-06,
"loss": 0.5460361838340759,
"step": 1353
},
{
"epoch": 1.629741119807345,
"grad_norm": 0.7310242255012503,
"learning_rate": 9.166486016280562e-06,
"loss": 0.6130272150039673,
"step": 1354
},
{
"epoch": 1.6309452137266707,
"grad_norm": 0.6884079103542232,
"learning_rate": 9.109119974936214e-06,
"loss": 0.6136751174926758,
"step": 1355
},
{
"epoch": 1.6321493076459963,
"grad_norm": 0.6529933620631292,
"learning_rate": 9.051916007110734e-06,
"loss": 0.6048874855041504,
"step": 1356
},
{
"epoch": 1.633353401565322,
"grad_norm": 0.6395968359912824,
"learning_rate": 8.994874339535569e-06,
"loss": 0.5344066023826599,
"step": 1357
},
{
"epoch": 1.6345574954846478,
"grad_norm": 0.6918279947344644,
"learning_rate": 8.937995198298876e-06,
"loss": 0.5948113799095154,
"step": 1358
},
{
"epoch": 1.6357615894039736,
"grad_norm": 0.749627446048807,
"learning_rate": 8.881278808844634e-06,
"loss": 0.7015585899353027,
"step": 1359
},
{
"epoch": 1.6369656833232993,
"grad_norm": 0.6513184099746449,
"learning_rate": 8.824725395971745e-06,
"loss": 0.5625496506690979,
"step": 1360
},
{
"epoch": 1.638169777242625,
"grad_norm": 0.70238524053711,
"learning_rate": 8.768335183833132e-06,
"loss": 0.5731056928634644,
"step": 1361
},
{
"epoch": 1.6393738711619505,
"grad_norm": 0.5922414879341457,
"learning_rate": 8.712108395934881e-06,
"loss": 0.5841649174690247,
"step": 1362
},
{
"epoch": 1.6405779650812762,
"grad_norm": 0.7120993016869709,
"learning_rate": 8.656045255135314e-06,
"loss": 0.6364700794219971,
"step": 1363
},
{
"epoch": 1.641782059000602,
"grad_norm": 0.6823205060765781,
"learning_rate": 8.600145983644148e-06,
"loss": 0.5725845694541931,
"step": 1364
},
{
"epoch": 1.642986152919928,
"grad_norm": 0.7198857587229472,
"learning_rate": 8.544410803021574e-06,
"loss": 0.6214855909347534,
"step": 1365
},
{
"epoch": 1.6441902468392535,
"grad_norm": 0.7599273701755161,
"learning_rate": 8.488839934177422e-06,
"loss": 0.5974388122558594,
"step": 1366
},
{
"epoch": 1.6453943407585792,
"grad_norm": 0.6299613926355601,
"learning_rate": 8.43343359737025e-06,
"loss": 0.541568398475647,
"step": 1367
},
{
"epoch": 1.6465984346779048,
"grad_norm": 0.7683368730630007,
"learning_rate": 8.37819201220647e-06,
"loss": 0.5598741173744202,
"step": 1368
},
{
"epoch": 1.6478025285972304,
"grad_norm": 0.6726052056551877,
"learning_rate": 8.323115397639513e-06,
"loss": 0.6542848348617554,
"step": 1369
},
{
"epoch": 1.6490066225165563,
"grad_norm": 0.7069030163623019,
"learning_rate": 8.268203971968924e-06,
"loss": 0.6667160987854004,
"step": 1370
},
{
"epoch": 1.6502107164358821,
"grad_norm": 0.7887578536010386,
"learning_rate": 8.213457952839527e-06,
"loss": 0.6406658887863159,
"step": 1371
},
{
"epoch": 1.6514148103552078,
"grad_norm": 0.6428228995877527,
"learning_rate": 8.158877557240529e-06,
"loss": 0.5842369794845581,
"step": 1372
},
{
"epoch": 1.6526189042745334,
"grad_norm": 0.6479207628949376,
"learning_rate": 8.104463001504697e-06,
"loss": 0.6519626379013062,
"step": 1373
},
{
"epoch": 1.653822998193859,
"grad_norm": 0.646271984786154,
"learning_rate": 8.05021450130744e-06,
"loss": 0.672814667224884,
"step": 1374
},
{
"epoch": 1.6550270921131849,
"grad_norm": 0.690089646787334,
"learning_rate": 7.996132271666062e-06,
"loss": 0.6398512125015259,
"step": 1375
},
{
"epoch": 1.6562311860325105,
"grad_norm": 0.7142843238137437,
"learning_rate": 7.942216526938795e-06,
"loss": 0.6502315998077393,
"step": 1376
},
{
"epoch": 1.6574352799518364,
"grad_norm": 0.7017333242505988,
"learning_rate": 7.888467480824003e-06,
"loss": 0.5540474653244019,
"step": 1377
},
{
"epoch": 1.658639373871162,
"grad_norm": 0.6112300966080016,
"learning_rate": 7.83488534635936e-06,
"loss": 0.5615662336349487,
"step": 1378
},
{
"epoch": 1.6598434677904876,
"grad_norm": 0.6239463125390158,
"learning_rate": 7.78147033592091e-06,
"loss": 0.5031113028526306,
"step": 1379
},
{
"epoch": 1.6610475617098133,
"grad_norm": 0.6256029475727574,
"learning_rate": 7.728222661222372e-06,
"loss": 0.5563163161277771,
"step": 1380
},
{
"epoch": 1.6622516556291391,
"grad_norm": 0.6174721327715081,
"learning_rate": 7.675142533314172e-06,
"loss": 0.5445525050163269,
"step": 1381
},
{
"epoch": 1.6634557495484648,
"grad_norm": 0.7024225984065051,
"learning_rate": 7.622230162582656e-06,
"loss": 0.645501971244812,
"step": 1382
},
{
"epoch": 1.6646598434677906,
"grad_norm": 0.669124661367454,
"learning_rate": 7.569485758749262e-06,
"loss": 0.591521680355072,
"step": 1383
},
{
"epoch": 1.6658639373871162,
"grad_norm": 0.6527263767067266,
"learning_rate": 7.5169095308696865e-06,
"loss": 0.5606772899627686,
"step": 1384
},
{
"epoch": 1.6670680313064419,
"grad_norm": 0.7128844407615076,
"learning_rate": 7.4645016873330366e-06,
"loss": 0.5577992796897888,
"step": 1385
},
{
"epoch": 1.6682721252257675,
"grad_norm": 0.7093763173916899,
"learning_rate": 7.412262435861023e-06,
"loss": 0.6242444515228271,
"step": 1386
},
{
"epoch": 1.6694762191450934,
"grad_norm": 0.636491032457578,
"learning_rate": 7.360191983507153e-06,
"loss": 0.6403906345367432,
"step": 1387
},
{
"epoch": 1.670680313064419,
"grad_norm": 0.6790989386676121,
"learning_rate": 7.308290536655832e-06,
"loss": 0.5881612300872803,
"step": 1388
},
{
"epoch": 1.6718844069837449,
"grad_norm": 0.6914623079540592,
"learning_rate": 7.256558301021665e-06,
"loss": 0.6011528968811035,
"step": 1389
},
{
"epoch": 1.6730885009030705,
"grad_norm": 0.6449236469256695,
"learning_rate": 7.2049954816485465e-06,
"loss": 0.5145377516746521,
"step": 1390
},
{
"epoch": 1.6742925948223961,
"grad_norm": 0.6402647292844768,
"learning_rate": 7.153602282908877e-06,
"loss": 0.46896347403526306,
"step": 1391
},
{
"epoch": 1.6754966887417218,
"grad_norm": 0.9044826757800251,
"learning_rate": 7.102378908502783e-06,
"loss": 0.5730105638504028,
"step": 1392
},
{
"epoch": 1.6767007826610476,
"grad_norm": 0.6755349959326966,
"learning_rate": 7.051325561457217e-06,
"loss": 0.49916258454322815,
"step": 1393
},
{
"epoch": 1.6779048765803732,
"grad_norm": 0.6081660544641575,
"learning_rate": 7.000442444125299e-06,
"loss": 0.581791877746582,
"step": 1394
},
{
"epoch": 1.679108970499699,
"grad_norm": 0.6568587395069151,
"learning_rate": 6.949729758185386e-06,
"loss": 0.5793961882591248,
"step": 1395
},
{
"epoch": 1.6803130644190247,
"grad_norm": 0.6045398063594519,
"learning_rate": 6.899187704640325e-06,
"loss": 0.5350834131240845,
"step": 1396
},
{
"epoch": 1.6815171583383504,
"grad_norm": 0.703670828010958,
"learning_rate": 6.848816483816672e-06,
"loss": 0.6520372629165649,
"step": 1397
},
{
"epoch": 1.682721252257676,
"grad_norm": 0.6930075174428402,
"learning_rate": 6.798616295363819e-06,
"loss": 0.6430896520614624,
"step": 1398
},
{
"epoch": 1.6839253461770018,
"grad_norm": 0.7730269428876103,
"learning_rate": 6.748587338253337e-06,
"loss": 0.6163848638534546,
"step": 1399
},
{
"epoch": 1.6851294400963275,
"grad_norm": 0.6286066574174697,
"learning_rate": 6.698729810778065e-06,
"loss": 0.5772340297698975,
"step": 1400
},
{
"epoch": 1.6863335340156533,
"grad_norm": 0.7294910037496207,
"learning_rate": 6.649043910551394e-06,
"loss": 0.5512592792510986,
"step": 1401
},
{
"epoch": 1.687537627934979,
"grad_norm": 0.6244222286619725,
"learning_rate": 6.59952983450643e-06,
"loss": 0.6313691139221191,
"step": 1402
},
{
"epoch": 1.6887417218543046,
"grad_norm": 0.6734978231377051,
"learning_rate": 6.550187778895245e-06,
"loss": 0.5909968614578247,
"step": 1403
},
{
"epoch": 1.6899458157736302,
"grad_norm": 0.6618866839532577,
"learning_rate": 6.501017939288145e-06,
"loss": 0.5310465693473816,
"step": 1404
},
{
"epoch": 1.691149909692956,
"grad_norm": 0.8607611105372066,
"learning_rate": 6.452020510572798e-06,
"loss": 0.6151556968688965,
"step": 1405
},
{
"epoch": 1.6923540036122817,
"grad_norm": 0.7005150491998585,
"learning_rate": 6.4031956869535446e-06,
"loss": 0.5600522756576538,
"step": 1406
},
{
"epoch": 1.6935580975316076,
"grad_norm": 0.6913763418991744,
"learning_rate": 6.354543661950552e-06,
"loss": 0.6204140186309814,
"step": 1407
},
{
"epoch": 1.6947621914509332,
"grad_norm": 0.6717318012350875,
"learning_rate": 6.3060646283991106e-06,
"loss": 0.5715824365615845,
"step": 1408
},
{
"epoch": 1.6959662853702588,
"grad_norm": 0.6804374521308102,
"learning_rate": 6.257758778448886e-06,
"loss": 0.6015537977218628,
"step": 1409
},
{
"epoch": 1.6971703792895845,
"grad_norm": 0.7197309714970597,
"learning_rate": 6.20962630356306e-06,
"loss": 0.6230743527412415,
"step": 1410
},
{
"epoch": 1.6983744732089103,
"grad_norm": 0.6084567912170616,
"learning_rate": 6.1616673945176836e-06,
"loss": 0.5166609287261963,
"step": 1411
},
{
"epoch": 1.699578567128236,
"grad_norm": 0.7166780063988029,
"learning_rate": 6.113882241400804e-06,
"loss": 0.5911763310432434,
"step": 1412
},
{
"epoch": 1.7007826610475618,
"grad_norm": 0.680644684252157,
"learning_rate": 6.0662710336118425e-06,
"loss": 0.5495004653930664,
"step": 1413
},
{
"epoch": 1.7019867549668874,
"grad_norm": 0.6541568159495065,
"learning_rate": 6.018833959860753e-06,
"loss": 0.5092551708221436,
"step": 1414
},
{
"epoch": 1.703190848886213,
"grad_norm": 0.7016217840045879,
"learning_rate": 5.971571208167298e-06,
"loss": 0.5613583326339722,
"step": 1415
},
{
"epoch": 1.7043949428055387,
"grad_norm": 0.6856085040176679,
"learning_rate": 5.9244829658602845e-06,
"loss": 0.5862411260604858,
"step": 1416
},
{
"epoch": 1.7055990367248646,
"grad_norm": 0.6591642311688184,
"learning_rate": 5.87756941957685e-06,
"loss": 0.5513682961463928,
"step": 1417
},
{
"epoch": 1.7068031306441902,
"grad_norm": 0.7349519639802647,
"learning_rate": 5.830830755261751e-06,
"loss": 0.6334446668624878,
"step": 1418
},
{
"epoch": 1.708007224563516,
"grad_norm": 0.73882863177204,
"learning_rate": 5.784267158166551e-06,
"loss": 0.6603875160217285,
"step": 1419
},
{
"epoch": 1.7092113184828417,
"grad_norm": 0.7095238297236984,
"learning_rate": 5.737878812848929e-06,
"loss": 0.6795850396156311,
"step": 1420
},
{
"epoch": 1.7104154124021673,
"grad_norm": 0.6563617435177955,
"learning_rate": 5.691665903171939e-06,
"loss": 0.5577661991119385,
"step": 1421
},
{
"epoch": 1.711619506321493,
"grad_norm": 0.6926805845112587,
"learning_rate": 5.6456286123032884e-06,
"loss": 0.5532077550888062,
"step": 1422
},
{
"epoch": 1.7128236002408188,
"grad_norm": 0.757840187820166,
"learning_rate": 5.599767122714627e-06,
"loss": 0.6900238394737244,
"step": 1423
},
{
"epoch": 1.7140276941601444,
"grad_norm": 0.6736091606976441,
"learning_rate": 5.554081616180801e-06,
"loss": 0.671580970287323,
"step": 1424
},
{
"epoch": 1.7152317880794703,
"grad_norm": 0.8110605807535137,
"learning_rate": 5.508572273779089e-06,
"loss": 0.6170952916145325,
"step": 1425
},
{
"epoch": 1.716435881998796,
"grad_norm": 0.801663317237608,
"learning_rate": 5.4632392758885985e-06,
"loss": 0.6166957020759583,
"step": 1426
},
{
"epoch": 1.7176399759181216,
"grad_norm": 0.7132870889967348,
"learning_rate": 5.418082802189434e-06,
"loss": 0.6364924907684326,
"step": 1427
},
{
"epoch": 1.7188440698374472,
"grad_norm": 0.6382040908202296,
"learning_rate": 5.37310303166208e-06,
"loss": 0.5974953174591064,
"step": 1428
},
{
"epoch": 1.720048163756773,
"grad_norm": 0.6293311603232586,
"learning_rate": 5.328300142586629e-06,
"loss": 0.5622908473014832,
"step": 1429
},
{
"epoch": 1.7212522576760987,
"grad_norm": 0.7087979265244937,
"learning_rate": 5.283674312542064e-06,
"loss": 0.6232190132141113,
"step": 1430
},
{
"epoch": 1.7224563515954245,
"grad_norm": 0.6819379024095552,
"learning_rate": 5.239225718405616e-06,
"loss": 0.5627528429031372,
"step": 1431
},
{
"epoch": 1.7236604455147502,
"grad_norm": 0.6636651792900646,
"learning_rate": 5.194954536352021e-06,
"loss": 0.5178531408309937,
"step": 1432
},
{
"epoch": 1.7248645394340758,
"grad_norm": 0.6221452686560454,
"learning_rate": 5.150860941852842e-06,
"loss": 0.5021407604217529,
"step": 1433
},
{
"epoch": 1.7260686333534014,
"grad_norm": 0.7076986193166933,
"learning_rate": 5.106945109675759e-06,
"loss": 0.5684460997581482,
"step": 1434
},
{
"epoch": 1.7272727272727273,
"grad_norm": 0.7586911684097273,
"learning_rate": 5.0632072138838584e-06,
"loss": 0.622063159942627,
"step": 1435
},
{
"epoch": 1.728476821192053,
"grad_norm": 0.6432313034645677,
"learning_rate": 5.019647427834978e-06,
"loss": 0.5491294860839844,
"step": 1436
},
{
"epoch": 1.7296809151113788,
"grad_norm": 0.7126328051936873,
"learning_rate": 4.9762659241810285e-06,
"loss": 0.6418676376342773,
"step": 1437
},
{
"epoch": 1.7308850090307044,
"grad_norm": 0.7245289716767404,
"learning_rate": 4.933062874867267e-06,
"loss": 0.6219337582588196,
"step": 1438
},
{
"epoch": 1.73208910295003,
"grad_norm": 0.6652968238429772,
"learning_rate": 4.8900384511316234e-06,
"loss": 0.5271619558334351,
"step": 1439
},
{
"epoch": 1.7332931968693557,
"grad_norm": 0.6528164827219402,
"learning_rate": 4.847192823504054e-06,
"loss": 0.5417010188102722,
"step": 1440
},
{
"epoch": 1.7344972907886815,
"grad_norm": 0.6721322436180379,
"learning_rate": 4.804526161805833e-06,
"loss": 0.4796888530254364,
"step": 1441
},
{
"epoch": 1.7357013847080074,
"grad_norm": 0.6753296354279802,
"learning_rate": 4.7620386351489185e-06,
"loss": 0.5415114164352417,
"step": 1442
},
{
"epoch": 1.736905478627333,
"grad_norm": 0.8239056192797215,
"learning_rate": 4.7197304119352355e-06,
"loss": 0.6290271282196045,
"step": 1443
},
{
"epoch": 1.7381095725466587,
"grad_norm": 0.7452282782525684,
"learning_rate": 4.6776016598560124e-06,
"loss": 0.5112418532371521,
"step": 1444
},
{
"epoch": 1.7393136664659843,
"grad_norm": 0.8020314546802029,
"learning_rate": 4.635652545891156e-06,
"loss": 0.5837382674217224,
"step": 1445
},
{
"epoch": 1.74051776038531,
"grad_norm": 0.6592594691489769,
"learning_rate": 4.5938832363085495e-06,
"loss": 0.5531149506568909,
"step": 1446
},
{
"epoch": 1.7417218543046358,
"grad_norm": 0.7155783650823891,
"learning_rate": 4.552293896663451e-06,
"loss": 0.5711277723312378,
"step": 1447
},
{
"epoch": 1.7429259482239616,
"grad_norm": 0.7090330366794828,
"learning_rate": 4.510884691797751e-06,
"loss": 0.6422768831253052,
"step": 1448
},
{
"epoch": 1.7441300421432873,
"grad_norm": 0.6767776010565582,
"learning_rate": 4.469655785839377e-06,
"loss": 0.5116891860961914,
"step": 1449
},
{
"epoch": 1.745334136062613,
"grad_norm": 0.7374313697530425,
"learning_rate": 4.428607342201635e-06,
"loss": 0.6454214453697205,
"step": 1450
},
{
"epoch": 1.7465382299819385,
"grad_norm": 0.7319555301101304,
"learning_rate": 4.387739523582551e-06,
"loss": 0.6378218531608582,
"step": 1451
},
{
"epoch": 1.7477423239012642,
"grad_norm": 0.6048272822670632,
"learning_rate": 4.347052491964265e-06,
"loss": 0.5626097321510315,
"step": 1452
},
{
"epoch": 1.74894641782059,
"grad_norm": 0.6527263143361477,
"learning_rate": 4.306546408612306e-06,
"loss": 0.5640212297439575,
"step": 1453
},
{
"epoch": 1.7501505117399159,
"grad_norm": 0.6880645867716209,
"learning_rate": 4.26622143407503e-06,
"loss": 0.5336100459098816,
"step": 1454
},
{
"epoch": 1.7513546056592415,
"grad_norm": 0.699862911543868,
"learning_rate": 4.226077728182959e-06,
"loss": 0.5081793069839478,
"step": 1455
},
{
"epoch": 1.7525586995785671,
"grad_norm": 0.7137368223623376,
"learning_rate": 4.186115450048128e-06,
"loss": 0.5804829001426697,
"step": 1456
},
{
"epoch": 1.7537627934978928,
"grad_norm": 0.7289237634290167,
"learning_rate": 4.1463347580635195e-06,
"loss": 0.5615112781524658,
"step": 1457
},
{
"epoch": 1.7549668874172184,
"grad_norm": 0.7339170280419827,
"learning_rate": 4.106735809902324e-06,
"loss": 0.5764937996864319,
"step": 1458
},
{
"epoch": 1.7561709813365443,
"grad_norm": 0.7654362997491899,
"learning_rate": 4.0673187625174195e-06,
"loss": 0.5667378902435303,
"step": 1459
},
{
"epoch": 1.7573750752558701,
"grad_norm": 0.7182102826832707,
"learning_rate": 4.028083772140689e-06,
"loss": 0.5884034633636475,
"step": 1460
},
{
"epoch": 1.7585791691751957,
"grad_norm": 0.6300590195339411,
"learning_rate": 3.989030994282434e-06,
"loss": 0.5312051773071289,
"step": 1461
},
{
"epoch": 1.7597832630945214,
"grad_norm": 0.6928354046394739,
"learning_rate": 3.950160583730761e-06,
"loss": 0.606346607208252,
"step": 1462
},
{
"epoch": 1.760987357013847,
"grad_norm": 0.6289904720451077,
"learning_rate": 3.911472694550916e-06,
"loss": 0.48723918199539185,
"step": 1463
},
{
"epoch": 1.7621914509331726,
"grad_norm": 0.6938945943340636,
"learning_rate": 3.872967480084727e-06,
"loss": 0.594094455242157,
"step": 1464
},
{
"epoch": 1.7633955448524985,
"grad_norm": 0.6454574999652682,
"learning_rate": 3.834645092949973e-06,
"loss": 0.5265861749649048,
"step": 1465
},
{
"epoch": 1.7645996387718244,
"grad_norm": 0.6402068002340231,
"learning_rate": 3.796505685039825e-06,
"loss": 0.5821572542190552,
"step": 1466
},
{
"epoch": 1.76580373269115,
"grad_norm": 0.6215539747644288,
"learning_rate": 3.758549407522144e-06,
"loss": 0.5699132680892944,
"step": 1467
},
{
"epoch": 1.7670078266104756,
"grad_norm": 0.6494984773034171,
"learning_rate": 3.720776410838983e-06,
"loss": 0.4552852511405945,
"step": 1468
},
{
"epoch": 1.7682119205298013,
"grad_norm": 0.7818456595599201,
"learning_rate": 3.6831868447059324e-06,
"loss": 0.5594058632850647,
"step": 1469
},
{
"epoch": 1.7694160144491269,
"grad_norm": 0.7190550070540783,
"learning_rate": 3.645780858111547e-06,
"loss": 0.5849020481109619,
"step": 1470
},
{
"epoch": 1.7706201083684527,
"grad_norm": 0.7249951662594688,
"learning_rate": 3.6085585993167805e-06,
"loss": 0.6334759593009949,
"step": 1471
},
{
"epoch": 1.7718242022877786,
"grad_norm": 0.67842444805327,
"learning_rate": 3.5715202158543125e-06,
"loss": 0.5594388246536255,
"step": 1472
},
{
"epoch": 1.7730282962071042,
"grad_norm": 0.8264073869124916,
"learning_rate": 3.5346658545280795e-06,
"loss": 0.6197201013565063,
"step": 1473
},
{
"epoch": 1.7742323901264299,
"grad_norm": 0.6698909050253031,
"learning_rate": 3.4979956614125953e-06,
"loss": 0.5784468650817871,
"step": 1474
},
{
"epoch": 1.7754364840457555,
"grad_norm": 0.7016937116133866,
"learning_rate": 3.4615097818524235e-06,
"loss": 0.5452237725257874,
"step": 1475
},
{
"epoch": 1.7766405779650811,
"grad_norm": 0.7086957002805386,
"learning_rate": 3.4252083604616182e-06,
"loss": 0.5750047564506531,
"step": 1476
},
{
"epoch": 1.777844671884407,
"grad_norm": 0.7512169881958669,
"learning_rate": 3.389091541123074e-06,
"loss": 0.6154974699020386,
"step": 1477
},
{
"epoch": 1.7790487658037328,
"grad_norm": 0.7042299855461226,
"learning_rate": 3.353159466988032e-06,
"loss": 0.5605722665786743,
"step": 1478
},
{
"epoch": 1.7802528597230585,
"grad_norm": 0.630558437887846,
"learning_rate": 3.3174122804754738e-06,
"loss": 0.5536834001541138,
"step": 1479
},
{
"epoch": 1.781456953642384,
"grad_norm": 0.6794569037867353,
"learning_rate": 3.2818501232715794e-06,
"loss": 0.5650051236152649,
"step": 1480
},
{
"epoch": 1.7826610475617097,
"grad_norm": 0.7591191616699784,
"learning_rate": 3.246473136329148e-06,
"loss": 0.51902836561203,
"step": 1481
},
{
"epoch": 1.7838651414810354,
"grad_norm": 0.707909160129663,
"learning_rate": 3.211281459867038e-06,
"loss": 0.5955489873886108,
"step": 1482
},
{
"epoch": 1.7850692354003612,
"grad_norm": 0.703632152706884,
"learning_rate": 3.1762752333696297e-06,
"loss": 0.5627739429473877,
"step": 1483
},
{
"epoch": 1.786273329319687,
"grad_norm": 0.7499834212374759,
"learning_rate": 3.141454595586252e-06,
"loss": 0.5941311717033386,
"step": 1484
},
{
"epoch": 1.7874774232390127,
"grad_norm": 0.7497386383473793,
"learning_rate": 3.1068196845306487e-06,
"loss": 0.6763837933540344,
"step": 1485
},
{
"epoch": 1.7886815171583383,
"grad_norm": 0.6661680797779713,
"learning_rate": 3.072370637480415e-06,
"loss": 0.5017702579498291,
"step": 1486
},
{
"epoch": 1.789885611077664,
"grad_norm": 0.6976050548874814,
"learning_rate": 3.0381075909764744e-06,
"loss": 0.5178176164627075,
"step": 1487
},
{
"epoch": 1.7910897049969896,
"grad_norm": 0.7283625311783152,
"learning_rate": 3.004030680822517e-06,
"loss": 0.6162016987800598,
"step": 1488
},
{
"epoch": 1.7922937989163155,
"grad_norm": 0.6842189977654847,
"learning_rate": 2.9701400420844737e-06,
"loss": 0.6359889507293701,
"step": 1489
},
{
"epoch": 1.7934978928356413,
"grad_norm": 0.6757745138342321,
"learning_rate": 2.9364358090899766e-06,
"loss": 0.6323922276496887,
"step": 1490
},
{
"epoch": 1.794701986754967,
"grad_norm": 0.6901477256781282,
"learning_rate": 2.9029181154278274e-06,
"loss": 0.6050222516059875,
"step": 1491
},
{
"epoch": 1.7959060806742926,
"grad_norm": 0.7171202358809439,
"learning_rate": 2.8695870939474624e-06,
"loss": 0.5529733896255493,
"step": 1492
},
{
"epoch": 1.7971101745936182,
"grad_norm": 0.6669003093248148,
"learning_rate": 2.836442876758438e-06,
"loss": 0.5076313018798828,
"step": 1493
},
{
"epoch": 1.7983142685129438,
"grad_norm": 0.7269113877004387,
"learning_rate": 2.8034855952299045e-06,
"loss": 0.5276248455047607,
"step": 1494
},
{
"epoch": 1.7995183624322697,
"grad_norm": 0.723355977685466,
"learning_rate": 2.770715379990069e-06,
"loss": 0.6064050793647766,
"step": 1495
},
{
"epoch": 1.8007224563515956,
"grad_norm": 0.8243187220777127,
"learning_rate": 2.7381323609256937e-06,
"loss": 0.5484981536865234,
"step": 1496
},
{
"epoch": 1.8019265502709212,
"grad_norm": 0.7039071786277653,
"learning_rate": 2.7057366671815856e-06,
"loss": 0.5383074283599854,
"step": 1497
},
{
"epoch": 1.8031306441902468,
"grad_norm": 0.6667856491817752,
"learning_rate": 2.6735284271600657e-06,
"loss": 0.547199547290802,
"step": 1498
},
{
"epoch": 1.8043347381095725,
"grad_norm": 0.6492279453493423,
"learning_rate": 2.641507768520479e-06,
"loss": 0.5876493453979492,
"step": 1499
},
{
"epoch": 1.805538832028898,
"grad_norm": 0.6305483892926103,
"learning_rate": 2.6096748181786758e-06,
"loss": 0.4917656183242798,
"step": 1500
},
{
"epoch": 1.806742925948224,
"grad_norm": 0.8067515837423501,
"learning_rate": 2.5780297023065057e-06,
"loss": 0.6041967272758484,
"step": 1501
},
{
"epoch": 1.8079470198675498,
"grad_norm": 0.6999590307341557,
"learning_rate": 2.546572546331338e-06,
"loss": 0.6243309378623962,
"step": 1502
},
{
"epoch": 1.8091511137868754,
"grad_norm": 0.6758862291871842,
"learning_rate": 2.5153034749355487e-06,
"loss": 0.5620877742767334,
"step": 1503
},
{
"epoch": 1.810355207706201,
"grad_norm": 0.7545413968501355,
"learning_rate": 2.4842226120560255e-06,
"loss": 0.6363770365715027,
"step": 1504
},
{
"epoch": 1.8115593016255267,
"grad_norm": 0.6875005723476368,
"learning_rate": 2.4533300808836757e-06,
"loss": 0.6246600151062012,
"step": 1505
},
{
"epoch": 1.8127633955448526,
"grad_norm": 0.6838453782037119,
"learning_rate": 2.4226260038629545e-06,
"loss": 0.5130318999290466,
"step": 1506
},
{
"epoch": 1.8139674894641782,
"grad_norm": 0.6661804546469481,
"learning_rate": 2.3921105026913527e-06,
"loss": 0.5225271582603455,
"step": 1507
},
{
"epoch": 1.815171583383504,
"grad_norm": 0.7559286846781317,
"learning_rate": 2.3617836983189366e-06,
"loss": 0.5840597152709961,
"step": 1508
},
{
"epoch": 1.8163756773028297,
"grad_norm": 0.7015898859383557,
"learning_rate": 2.3316457109478716e-06,
"loss": 0.576333224773407,
"step": 1509
},
{
"epoch": 1.8175797712221553,
"grad_norm": 0.7325334460929459,
"learning_rate": 2.3016966600319154e-06,
"loss": 0.6196116805076599,
"step": 1510
},
{
"epoch": 1.818783865141481,
"grad_norm": 0.7238971994565907,
"learning_rate": 2.2719366642759754e-06,
"loss": 0.6115279197692871,
"step": 1511
},
{
"epoch": 1.8199879590608068,
"grad_norm": 0.6615840220669031,
"learning_rate": 2.2423658416356296e-06,
"loss": 0.5739059448242188,
"step": 1512
},
{
"epoch": 1.8211920529801324,
"grad_norm": 0.6438818574320542,
"learning_rate": 2.212984309316646e-06,
"loss": 0.5544348955154419,
"step": 1513
},
{
"epoch": 1.8223961468994583,
"grad_norm": 0.6887199478226267,
"learning_rate": 2.183792183774541e-06,
"loss": 0.5506141781806946,
"step": 1514
},
{
"epoch": 1.823600240818784,
"grad_norm": 0.630566851328714,
"learning_rate": 2.1547895807141004e-06,
"loss": 0.5032554864883423,
"step": 1515
},
{
"epoch": 1.8248043347381095,
"grad_norm": 0.6613185082560529,
"learning_rate": 2.125976615088926e-06,
"loss": 0.6159868240356445,
"step": 1516
},
{
"epoch": 1.8260084286574352,
"grad_norm": 0.7007602126683692,
"learning_rate": 2.0973534011009823e-06,
"loss": 0.5271652936935425,
"step": 1517
},
{
"epoch": 1.827212522576761,
"grad_norm": 0.6991109886060091,
"learning_rate": 2.0689200522001294e-06,
"loss": 0.5943084955215454,
"step": 1518
},
{
"epoch": 1.8284166164960867,
"grad_norm": 0.6317388395635505,
"learning_rate": 2.040676681083703e-06,
"loss": 0.5195907354354858,
"step": 1519
},
{
"epoch": 1.8296207104154125,
"grad_norm": 0.6614371125543239,
"learning_rate": 2.01262339969604e-06,
"loss": 0.5257569551467896,
"step": 1520
},
{
"epoch": 1.8308248043347382,
"grad_norm": 0.6721043989031192,
"learning_rate": 1.9847603192280515e-06,
"loss": 0.5383394956588745,
"step": 1521
},
{
"epoch": 1.8320288982540638,
"grad_norm": 0.6516699678388052,
"learning_rate": 1.957087550116765e-06,
"loss": 0.6025786995887756,
"step": 1522
},
{
"epoch": 1.8332329921733894,
"grad_norm": 0.7086815397016576,
"learning_rate": 1.929605202044904e-06,
"loss": 0.624287486076355,
"step": 1523
},
{
"epoch": 1.8344370860927153,
"grad_norm": 0.7095880324856938,
"learning_rate": 1.9023133839404517e-06,
"loss": 0.609376847743988,
"step": 1524
},
{
"epoch": 1.835641180012041,
"grad_norm": 0.6952599672746529,
"learning_rate": 1.875212203976201e-06,
"loss": 0.578833281993866,
"step": 1525
},
{
"epoch": 1.8368452739313668,
"grad_norm": 0.6335110011108259,
"learning_rate": 1.8483017695693494e-06,
"loss": 0.49487796425819397,
"step": 1526
},
{
"epoch": 1.8380493678506924,
"grad_norm": 0.725572265875903,
"learning_rate": 1.8215821873810601e-06,
"loss": 0.644855260848999,
"step": 1527
},
{
"epoch": 1.839253461770018,
"grad_norm": 0.775538378700794,
"learning_rate": 1.7950535633160403e-06,
"loss": 0.5681218504905701,
"step": 1528
},
{
"epoch": 1.8404575556893437,
"grad_norm": 0.7179662032057308,
"learning_rate": 1.768716002522125e-06,
"loss": 0.5964992642402649,
"step": 1529
},
{
"epoch": 1.8416616496086695,
"grad_norm": 0.6403003850145351,
"learning_rate": 1.7425696093898548e-06,
"loss": 0.4977972209453583,
"step": 1530
},
{
"epoch": 1.8428657435279951,
"grad_norm": 0.7228884489273776,
"learning_rate": 1.7166144875520763e-06,
"loss": 0.5850366950035095,
"step": 1531
},
{
"epoch": 1.844069837447321,
"grad_norm": 0.695494118116519,
"learning_rate": 1.6908507398834927e-06,
"loss": 0.5370050668716431,
"step": 1532
},
{
"epoch": 1.8452739313666466,
"grad_norm": 0.6719266812533203,
"learning_rate": 1.6652784685003197e-06,
"loss": 0.5704593658447266,
"step": 1533
},
{
"epoch": 1.8464780252859723,
"grad_norm": 0.6905237547910387,
"learning_rate": 1.6398977747598243e-06,
"loss": 0.6007063984870911,
"step": 1534
},
{
"epoch": 1.847682119205298,
"grad_norm": 0.6724829526221116,
"learning_rate": 1.614708759259942e-06,
"loss": 0.5665451288223267,
"step": 1535
},
{
"epoch": 1.8488862131246238,
"grad_norm": 0.8393382950462466,
"learning_rate": 1.5897115218388936e-06,
"loss": 0.6753185987472534,
"step": 1536
},
{
"epoch": 1.8500903070439494,
"grad_norm": 0.6494036890210806,
"learning_rate": 1.564906161574764e-06,
"loss": 0.4877506494522095,
"step": 1537
},
{
"epoch": 1.8512944009632752,
"grad_norm": 0.68916513427791,
"learning_rate": 1.5402927767851239e-06,
"loss": 0.5553507208824158,
"step": 1538
},
{
"epoch": 1.8524984948826009,
"grad_norm": 0.723218474778968,
"learning_rate": 1.5158714650266414e-06,
"loss": 0.5621401071548462,
"step": 1539
},
{
"epoch": 1.8537025888019265,
"grad_norm": 0.7184256940054815,
"learning_rate": 1.4916423230946885e-06,
"loss": 0.5810825824737549,
"step": 1540
},
{
"epoch": 1.8549066827212521,
"grad_norm": 0.7104079844650215,
"learning_rate": 1.4676054470229517e-06,
"loss": 0.5943775177001953,
"step": 1541
},
{
"epoch": 1.856110776640578,
"grad_norm": 0.6415428432253708,
"learning_rate": 1.443760932083077e-06,
"loss": 0.5965542793273926,
"step": 1542
},
{
"epoch": 1.8573148705599036,
"grad_norm": 0.6686962536908114,
"learning_rate": 1.4201088727842648e-06,
"loss": 0.5232040882110596,
"step": 1543
},
{
"epoch": 1.8585189644792295,
"grad_norm": 0.7009469795092166,
"learning_rate": 1.3966493628729039e-06,
"loss": 0.6138120293617249,
"step": 1544
},
{
"epoch": 1.8597230583985551,
"grad_norm": 0.6208974700635488,
"learning_rate": 1.373382495332215e-06,
"loss": 0.49354854226112366,
"step": 1545
},
{
"epoch": 1.8609271523178808,
"grad_norm": 0.692130552786189,
"learning_rate": 1.3503083623818412e-06,
"loss": 0.6931569576263428,
"step": 1546
},
{
"epoch": 1.8621312462372064,
"grad_norm": 0.765254055365152,
"learning_rate": 1.3274270554775425e-06,
"loss": 0.7060236930847168,
"step": 1547
},
{
"epoch": 1.8633353401565322,
"grad_norm": 0.699732493479322,
"learning_rate": 1.3047386653107784e-06,
"loss": 0.606826663017273,
"step": 1548
},
{
"epoch": 1.8645394340758579,
"grad_norm": 0.6674193143847964,
"learning_rate": 1.282243281808393e-06,
"loss": 0.674689531326294,
"step": 1549
},
{
"epoch": 1.8657435279951837,
"grad_norm": 0.6577773378574601,
"learning_rate": 1.2599409941322081e-06,
"loss": 0.5309125781059265,
"step": 1550
},
{
"epoch": 1.8669476219145094,
"grad_norm": 0.7046706962824838,
"learning_rate": 1.2378318906787145e-06,
"loss": 0.5600446462631226,
"step": 1551
},
{
"epoch": 1.868151715833835,
"grad_norm": 0.6566519822968615,
"learning_rate": 1.2159160590787143e-06,
"loss": 0.5337668657302856,
"step": 1552
},
{
"epoch": 1.8693558097531606,
"grad_norm": 0.694918375362827,
"learning_rate": 1.1941935861969455e-06,
"loss": 0.573444664478302,
"step": 1553
},
{
"epoch": 1.8705599036724865,
"grad_norm": 0.6117497741372008,
"learning_rate": 1.1726645581317752e-06,
"loss": 0.5877199769020081,
"step": 1554
},
{
"epoch": 1.8717639975918121,
"grad_norm": 0.6849649765191942,
"learning_rate": 1.1513290602148174e-06,
"loss": 0.5103924870491028,
"step": 1555
},
{
"epoch": 1.872968091511138,
"grad_norm": 0.6666116162172622,
"learning_rate": 1.1301871770106332e-06,
"loss": 0.5621691942214966,
"step": 1556
},
{
"epoch": 1.8741721854304636,
"grad_norm": 0.6918792000850436,
"learning_rate": 1.1092389923163915e-06,
"loss": 0.585303783416748,
"step": 1557
},
{
"epoch": 1.8753762793497892,
"grad_norm": 0.7628439993607844,
"learning_rate": 1.0884845891614925e-06,
"loss": 0.6140367984771729,
"step": 1558
},
{
"epoch": 1.8765803732691149,
"grad_norm": 0.6740049850142918,
"learning_rate": 1.0679240498073118e-06,
"loss": 0.5484222173690796,
"step": 1559
},
{
"epoch": 1.8777844671884407,
"grad_norm": 0.7523385412885472,
"learning_rate": 1.0475574557467837e-06,
"loss": 0.5650358200073242,
"step": 1560
},
{
"epoch": 1.8789885611077664,
"grad_norm": 0.6880191801421335,
"learning_rate": 1.0273848877041802e-06,
"loss": 0.47756820917129517,
"step": 1561
},
{
"epoch": 1.8801926550270922,
"grad_norm": 0.7132237427539956,
"learning_rate": 1.0074064256347104e-06,
"loss": 0.5750943422317505,
"step": 1562
},
{
"epoch": 1.8813967489464178,
"grad_norm": 0.6463908501083647,
"learning_rate": 9.876221487242322e-07,
"loss": 0.5670018196105957,
"step": 1563
},
{
"epoch": 1.8826008428657435,
"grad_norm": 0.7603473365673563,
"learning_rate": 9.680321353889576e-07,
"loss": 0.5906625986099243,
"step": 1564
},
{
"epoch": 1.883804936785069,
"grad_norm": 0.6859465389776784,
"learning_rate": 9.486364632750878e-07,
"loss": 0.5674991607666016,
"step": 1565
},
{
"epoch": 1.885009030704395,
"grad_norm": 0.6769950863239539,
"learning_rate": 9.294352092585779e-07,
"loss": 0.5772174596786499,
"step": 1566
},
{
"epoch": 1.8862131246237208,
"grad_norm": 0.7155135900382078,
"learning_rate": 9.104284494447779e-07,
"loss": 0.5847698450088501,
"step": 1567
},
{
"epoch": 1.8874172185430464,
"grad_norm": 0.7238855876217561,
"learning_rate": 8.916162591681543e-07,
"loss": 0.5530298948287964,
"step": 1568
},
{
"epoch": 1.888621312462372,
"grad_norm": 0.6526045127217193,
"learning_rate": 8.729987129919682e-07,
"loss": 0.5043114423751831,
"step": 1569
},
{
"epoch": 1.8898254063816977,
"grad_norm": 0.7391424950916906,
"learning_rate": 8.545758847080143e-07,
"loss": 0.5303754806518555,
"step": 1570
},
{
"epoch": 1.8910295003010233,
"grad_norm": 0.6712333443189676,
"learning_rate": 8.363478473363107e-07,
"loss": 0.577226996421814,
"step": 1571
},
{
"epoch": 1.8922335942203492,
"grad_norm": 0.6981141582176953,
"learning_rate": 8.183146731247982e-07,
"loss": 0.5423426628112793,
"step": 1572
},
{
"epoch": 1.893437688139675,
"grad_norm": 0.6311937513785054,
"learning_rate": 8.004764335490856e-07,
"loss": 0.48133552074432373,
"step": 1573
},
{
"epoch": 1.8946417820590007,
"grad_norm": 0.6789891972188489,
"learning_rate": 7.828331993121163e-07,
"loss": 0.5657028555870056,
"step": 1574
},
{
"epoch": 1.8958458759783263,
"grad_norm": 0.7254998928602563,
"learning_rate": 7.653850403439411e-07,
"loss": 0.5331858396530151,
"step": 1575
},
{
"epoch": 1.897049969897652,
"grad_norm": 0.6608618601341222,
"learning_rate": 7.481320258014124e-07,
"loss": 0.5427488088607788,
"step": 1576
},
{
"epoch": 1.8982540638169776,
"grad_norm": 0.6694799327623477,
"learning_rate": 7.31074224067918e-07,
"loss": 0.6095342636108398,
"step": 1577
},
{
"epoch": 1.8994581577363034,
"grad_norm": 0.6099067398142338,
"learning_rate": 7.142117027531092e-07,
"loss": 0.5477356314659119,
"step": 1578
},
{
"epoch": 1.9006622516556293,
"grad_norm": 0.7085200907625167,
"learning_rate": 6.975445286926063e-07,
"loss": 0.5689785480499268,
"step": 1579
},
{
"epoch": 1.901866345574955,
"grad_norm": 0.7103499006587805,
"learning_rate": 6.810727679477935e-07,
"loss": 0.5828697681427002,
"step": 1580
},
{
"epoch": 1.9030704394942806,
"grad_norm": 0.6301829347830576,
"learning_rate": 6.647964858055133e-07,
"loss": 0.5717979669570923,
"step": 1581
},
{
"epoch": 1.9042745334136062,
"grad_norm": 0.636202610900024,
"learning_rate": 6.48715746777806e-07,
"loss": 0.5298157930374146,
"step": 1582
},
{
"epoch": 1.9054786273329318,
"grad_norm": 0.6411930882178205,
"learning_rate": 6.328306146016593e-07,
"loss": 0.4738658666610718,
"step": 1583
},
{
"epoch": 1.9066827212522577,
"grad_norm": 0.75869674875339,
"learning_rate": 6.171411522387871e-07,
"loss": 0.6281449794769287,
"step": 1584
},
{
"epoch": 1.9078868151715835,
"grad_norm": 0.6766625111559003,
"learning_rate": 6.016474218753288e-07,
"loss": 0.6420220136642456,
"step": 1585
},
{
"epoch": 1.9090909090909092,
"grad_norm": 0.6592870011567821,
"learning_rate": 5.863494849216444e-07,
"loss": 0.5731959342956543,
"step": 1586
},
{
"epoch": 1.9102950030102348,
"grad_norm": 0.7741711986152868,
"learning_rate": 5.712474020120484e-07,
"loss": 0.5774096250534058,
"step": 1587
},
{
"epoch": 1.9114990969295604,
"grad_norm": 0.7327164005422871,
"learning_rate": 5.563412330045758e-07,
"loss": 0.6661399602890015,
"step": 1588
},
{
"epoch": 1.912703190848886,
"grad_norm": 0.6490076102800001,
"learning_rate": 5.416310369807331e-07,
"loss": 0.5940079092979431,
"step": 1589
},
{
"epoch": 1.913907284768212,
"grad_norm": 0.6929070437022665,
"learning_rate": 5.271168722453035e-07,
"loss": 0.5624694228172302,
"step": 1590
},
{
"epoch": 1.9151113786875378,
"grad_norm": 0.7220541731385967,
"learning_rate": 5.127987963260583e-07,
"loss": 0.6027872562408447,
"step": 1591
},
{
"epoch": 1.9163154726068634,
"grad_norm": 0.7573965262363941,
"learning_rate": 4.986768659735852e-07,
"loss": 0.5407252907752991,
"step": 1592
},
{
"epoch": 1.917519566526189,
"grad_norm": 0.7140587243770241,
"learning_rate": 4.847511371610159e-07,
"loss": 0.5981488823890686,
"step": 1593
},
{
"epoch": 1.9187236604455147,
"grad_norm": 0.7092430794499894,
"learning_rate": 4.710216650838317e-07,
"loss": 0.6029732823371887,
"step": 1594
},
{
"epoch": 1.9199277543648403,
"grad_norm": 0.7233316220665095,
"learning_rate": 4.5748850415964774e-07,
"loss": 0.48785698413848877,
"step": 1595
},
{
"epoch": 1.9211318482841662,
"grad_norm": 0.7018765555724064,
"learning_rate": 4.4415170802797333e-07,
"loss": 0.627527117729187,
"step": 1596
},
{
"epoch": 1.922335942203492,
"grad_norm": 0.6960703296753732,
"learning_rate": 4.3101132955002396e-07,
"loss": 0.5867995023727417,
"step": 1597
},
{
"epoch": 1.9235400361228177,
"grad_norm": 0.7554260967626275,
"learning_rate": 4.180674208084989e-07,
"loss": 0.546118438243866,
"step": 1598
},
{
"epoch": 1.9247441300421433,
"grad_norm": 0.6780941213732067,
"learning_rate": 4.0532003310736475e-07,
"loss": 0.5627431273460388,
"step": 1599
},
{
"epoch": 1.925948223961469,
"grad_norm": 0.7461456750112884,
"learning_rate": 3.9276921697169455e-07,
"loss": 0.5974733829498291,
"step": 1600
},
{
"epoch": 1.9271523178807946,
"grad_norm": 0.7331210435274516,
"learning_rate": 3.804150221474179e-07,
"loss": 0.61258864402771,
"step": 1601
},
{
"epoch": 1.9283564118001204,
"grad_norm": 0.7291842214281149,
"learning_rate": 3.6825749760113215e-07,
"loss": 0.648840606212616,
"step": 1602
},
{
"epoch": 1.9295605057194463,
"grad_norm": 0.7458366591965205,
"learning_rate": 3.5629669151994725e-07,
"loss": 0.6015305519104004,
"step": 1603
},
{
"epoch": 1.930764599638772,
"grad_norm": 0.8293947514840635,
"learning_rate": 3.4453265131124677e-07,
"loss": 0.5498949289321899,
"step": 1604
},
{
"epoch": 1.9319686935580975,
"grad_norm": 0.6484978877567261,
"learning_rate": 3.3296542360253256e-07,
"loss": 0.5641001462936401,
"step": 1605
},
{
"epoch": 1.9331727874774232,
"grad_norm": 0.6847140731160282,
"learning_rate": 3.2159505424122495e-07,
"loss": 0.5579475164413452,
"step": 1606
},
{
"epoch": 1.9343768813967488,
"grad_norm": 0.7236561832059804,
"learning_rate": 3.1042158829447385e-07,
"loss": 0.6008574962615967,
"step": 1607
},
{
"epoch": 1.9355809753160746,
"grad_norm": 0.6680008850614204,
"learning_rate": 2.9944507004900367e-07,
"loss": 0.5793865919113159,
"step": 1608
},
{
"epoch": 1.9367850692354005,
"grad_norm": 0.7226863078514579,
"learning_rate": 2.8866554301091866e-07,
"loss": 0.5802816152572632,
"step": 1609
},
{
"epoch": 1.9379891631547261,
"grad_norm": 0.6066089323199368,
"learning_rate": 2.780830499055476e-07,
"loss": 0.41039422154426575,
"step": 1610
},
{
"epoch": 1.9391932570740518,
"grad_norm": 0.7541554507302138,
"learning_rate": 2.6769763267723845e-07,
"loss": 0.7123790979385376,
"step": 1611
},
{
"epoch": 1.9403973509933774,
"grad_norm": 0.6539368475453213,
"learning_rate": 2.575093324892364e-07,
"loss": 0.5050459504127502,
"step": 1612
},
{
"epoch": 1.941601444912703,
"grad_norm": 0.6362805423411522,
"learning_rate": 2.4751818972350016e-07,
"loss": 0.540932297706604,
"step": 1613
},
{
"epoch": 1.9428055388320289,
"grad_norm": 0.7530210700051205,
"learning_rate": 2.377242439805305e-07,
"loss": 0.7373020052909851,
"step": 1614
},
{
"epoch": 1.9440096327513547,
"grad_norm": 0.7606614768716228,
"learning_rate": 2.281275340792477e-07,
"loss": 0.6150696277618408,
"step": 1615
},
{
"epoch": 1.9452137266706804,
"grad_norm": 0.6667905800819639,
"learning_rate": 2.187280980567863e-07,
"loss": 0.6177890300750732,
"step": 1616
},
{
"epoch": 1.946417820590006,
"grad_norm": 0.7280364011212537,
"learning_rate": 2.095259731684007e-07,
"loss": 0.535525381565094,
"step": 1617
},
{
"epoch": 1.9476219145093316,
"grad_norm": 0.6612643727912537,
"learning_rate": 2.0052119588727103e-07,
"loss": 0.5701784491539001,
"step": 1618
},
{
"epoch": 1.9488260084286573,
"grad_norm": 0.6930350525511971,
"learning_rate": 1.917138019043918e-07,
"loss": 0.5127325057983398,
"step": 1619
},
{
"epoch": 1.9500301023479831,
"grad_norm": 0.6645178871748494,
"learning_rate": 1.8310382612841125e-07,
"loss": 0.5932685136795044,
"step": 1620
},
{
"epoch": 1.951234196267309,
"grad_norm": 0.6600103793645039,
"learning_rate": 1.7469130268549238e-07,
"loss": 0.5705520510673523,
"step": 1621
},
{
"epoch": 1.9524382901866346,
"grad_norm": 0.7037195958313917,
"learning_rate": 1.6647626491919088e-07,
"loss": 0.6251012086868286,
"step": 1622
},
{
"epoch": 1.9536423841059603,
"grad_norm": 0.7203644868401232,
"learning_rate": 1.5845874539032192e-07,
"loss": 0.6126412153244019,
"step": 1623
},
{
"epoch": 1.9548464780252859,
"grad_norm": 0.6893511239479428,
"learning_rate": 1.5063877587681019e-07,
"loss": 0.6284942626953125,
"step": 1624
},
{
"epoch": 1.9560505719446115,
"grad_norm": 0.7277088854826446,
"learning_rate": 1.4301638737358459e-07,
"loss": 0.5494322776794434,
"step": 1625
},
{
"epoch": 1.9572546658639374,
"grad_norm": 0.7044708997407658,
"learning_rate": 1.3559161009246146e-07,
"loss": 0.4695800542831421,
"step": 1626
},
{
"epoch": 1.9584587597832632,
"grad_norm": 0.6464366555921628,
"learning_rate": 1.283644734619893e-07,
"loss": 0.5801255106925964,
"step": 1627
},
{
"epoch": 1.9596628537025889,
"grad_norm": 0.6572737348994054,
"learning_rate": 1.2133500612737103e-07,
"loss": 0.5532472133636475,
"step": 1628
},
{
"epoch": 1.9608669476219145,
"grad_norm": 0.6546752224566054,
"learning_rate": 1.1450323595034174e-07,
"loss": 0.5728014707565308,
"step": 1629
},
{
"epoch": 1.9620710415412401,
"grad_norm": 0.7098695274037261,
"learning_rate": 1.0786919000903562e-07,
"loss": 0.6312659978866577,
"step": 1630
},
{
"epoch": 1.963275135460566,
"grad_norm": 0.6431445514270803,
"learning_rate": 1.0143289459790816e-07,
"loss": 0.5799425840377808,
"step": 1631
},
{
"epoch": 1.9644792293798916,
"grad_norm": 0.6719495518124794,
"learning_rate": 9.519437522760299e-08,
"loss": 0.5714237689971924,
"step": 1632
},
{
"epoch": 1.9656833232992175,
"grad_norm": 0.752533416502612,
"learning_rate": 8.915365662488518e-08,
"loss": 0.5689437985420227,
"step": 1633
},
{
"epoch": 1.966887417218543,
"grad_norm": 0.7692727283179666,
"learning_rate": 8.331076273250249e-08,
"loss": 0.6392886638641357,
"step": 1634
},
{
"epoch": 1.9680915111378687,
"grad_norm": 0.7336589415290271,
"learning_rate": 7.766571670913547e-08,
"loss": 0.5719316005706787,
"step": 1635
},
{
"epoch": 1.9692956050571944,
"grad_norm": 0.6838298911100253,
"learning_rate": 7.221854092926971e-08,
"loss": 0.5827243328094482,
"step": 1636
},
{
"epoch": 1.9704996989765202,
"grad_norm": 0.7686224124765682,
"learning_rate": 6.696925698311817e-08,
"loss": 0.6425184011459351,
"step": 1637
},
{
"epoch": 1.9717037928958459,
"grad_norm": 0.6532559866884581,
"learning_rate": 6.191788567654344e-08,
"loss": 0.5946167707443237,
"step": 1638
},
{
"epoch": 1.9729078868151717,
"grad_norm": 0.6312344261575024,
"learning_rate": 5.706444703096336e-08,
"loss": 0.5090816617012024,
"step": 1639
},
{
"epoch": 1.9741119807344973,
"grad_norm": 0.7676146358721873,
"learning_rate": 5.240896028327891e-08,
"loss": 0.579215407371521,
"step": 1640
},
{
"epoch": 1.975316074653823,
"grad_norm": 0.6728133395765841,
"learning_rate": 4.7951443885807524e-08,
"loss": 0.5652763247489929,
"step": 1641
},
{
"epoch": 1.9765201685731486,
"grad_norm": 0.6432580000680586,
"learning_rate": 4.3691915506177686e-08,
"loss": 0.5408779382705688,
"step": 1642
},
{
"epoch": 1.9777242624924745,
"grad_norm": 0.6315126229459764,
"learning_rate": 3.963039202730112e-08,
"loss": 0.5882087349891663,
"step": 1643
},
{
"epoch": 1.9789283564118,
"grad_norm": 0.7011158641139948,
"learning_rate": 3.576688954727847e-08,
"loss": 0.5176501274108887,
"step": 1644
},
{
"epoch": 1.980132450331126,
"grad_norm": 0.6587097052056431,
"learning_rate": 3.210142337932709e-08,
"loss": 0.5108386874198914,
"step": 1645
},
{
"epoch": 1.9813365442504516,
"grad_norm": 0.6425909595737344,
"learning_rate": 2.8634008051758864e-08,
"loss": 0.5777048468589783,
"step": 1646
},
{
"epoch": 1.9825406381697772,
"grad_norm": 0.7147004863851809,
"learning_rate": 2.536465730788029e-08,
"loss": 0.500884473323822,
"step": 1647
},
{
"epoch": 1.9837447320891028,
"grad_norm": 0.6737989717991144,
"learning_rate": 2.229338410597026e-08,
"loss": 0.5754954218864441,
"step": 1648
},
{
"epoch": 1.9849488260084287,
"grad_norm": 0.7131714825607306,
"learning_rate": 1.9420200619207907e-08,
"loss": 0.5816112160682678,
"step": 1649
},
{
"epoch": 1.9861529199277543,
"grad_norm": 0.7554088529786651,
"learning_rate": 1.6745118235628188e-08,
"loss": 0.6610931158065796,
"step": 1650
},
{
"epoch": 1.9873570138470802,
"grad_norm": 0.7420931407483101,
"learning_rate": 1.4268147558088585e-08,
"loss": 0.5348343849182129,
"step": 1651
},
{
"epoch": 1.9885611077664058,
"grad_norm": 0.7690940969965758,
"learning_rate": 1.1989298404213588e-08,
"loss": 0.5157413482666016,
"step": 1652
},
{
"epoch": 1.9897652016857315,
"grad_norm": 0.6882882292897665,
"learning_rate": 9.908579806361396e-09,
"loss": 0.4572542905807495,
"step": 1653
},
{
"epoch": 1.990969295605057,
"grad_norm": 0.6899079935771424,
"learning_rate": 8.026000011596146e-09,
"loss": 0.6765772700309753,
"step": 1654
},
{
"epoch": 1.992173389524383,
"grad_norm": 0.6843041221046693,
"learning_rate": 6.341566481626871e-09,
"loss": 0.6102558374404907,
"step": 1655
},
{
"epoch": 1.9933774834437086,
"grad_norm": 0.6850570474611285,
"learning_rate": 4.855285892813033e-09,
"loss": 0.5829460024833679,
"step": 1656
},
{
"epoch": 1.9945815773630344,
"grad_norm": 0.6410933816390361,
"learning_rate": 3.567164136120127e-09,
"loss": 0.5028737187385559,
"step": 1657
},
{
"epoch": 1.99578567128236,
"grad_norm": 0.65338875182292,
"learning_rate": 2.47720631710302e-09,
"loss": 0.5529558658599854,
"step": 1658
},
{
"epoch": 1.9969897652016857,
"grad_norm": 0.6618217221632864,
"learning_rate": 1.5854167558670975e-09,
"loss": 0.5901626348495483,
"step": 1659
},
{
"epoch": 1.9981938591210113,
"grad_norm": 0.6620340323637777,
"learning_rate": 8.917989870849131e-10,
"loss": 0.55892014503479,
"step": 1660
},
{
"epoch": 1.9993979530403372,
"grad_norm": 0.7584505389100632,
"learning_rate": 3.9635575994623196e-10,
"loss": 0.613528311252594,
"step": 1661
},
{
"epoch": 2.0,
"grad_norm": 0.897855344686248,
"learning_rate": 9.908903817468229e-11,
"loss": 0.5595951080322266,
"step": 1662
},
{
"epoch": 2.0,
"step": 1662,
"total_flos": 1137613352534016.0,
"train_loss": 0.6870017016730153,
"train_runtime": 15224.812,
"train_samples_per_second": 0.873,
"train_steps_per_second": 0.109
}
],
"logging_steps": 1,
"max_steps": 1662,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 100,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 1137613352534016.0,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}