gyung's picture
Add files using upload-large-folder tool
002ea5f verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"eval_steps": 500,
"global_step": 1460,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0013698630136986301,
"grad_norm": 0.9543317556381226,
"learning_rate": 0.0,
"loss": 1.289506435394287,
"step": 1
},
{
"epoch": 0.0027397260273972603,
"grad_norm": 1.102232575416565,
"learning_rate": 4.5454545454545457e-07,
"loss": 1.419710397720337,
"step": 2
},
{
"epoch": 0.00410958904109589,
"grad_norm": 1.055787205696106,
"learning_rate": 9.090909090909091e-07,
"loss": 1.2828402519226074,
"step": 3
},
{
"epoch": 0.005479452054794521,
"grad_norm": 1.1042431592941284,
"learning_rate": 1.3636363636363636e-06,
"loss": 1.3904743194580078,
"step": 4
},
{
"epoch": 0.00684931506849315,
"grad_norm": 0.9779807925224304,
"learning_rate": 1.8181818181818183e-06,
"loss": 1.307753086090088,
"step": 5
},
{
"epoch": 0.00821917808219178,
"grad_norm": 0.9276996850967407,
"learning_rate": 2.2727272727272728e-06,
"loss": 1.3470945358276367,
"step": 6
},
{
"epoch": 0.009589041095890411,
"grad_norm": 0.8735243678092957,
"learning_rate": 2.7272727272727272e-06,
"loss": 1.2755506038665771,
"step": 7
},
{
"epoch": 0.010958904109589041,
"grad_norm": 0.7032255530357361,
"learning_rate": 3.181818181818182e-06,
"loss": 1.2992908954620361,
"step": 8
},
{
"epoch": 0.012328767123287671,
"grad_norm": 0.6715389490127563,
"learning_rate": 3.6363636363636366e-06,
"loss": 1.2155871391296387,
"step": 9
},
{
"epoch": 0.0136986301369863,
"grad_norm": 0.5294431447982788,
"learning_rate": 4.0909090909090915e-06,
"loss": 1.0370020866394043,
"step": 10
},
{
"epoch": 0.015068493150684932,
"grad_norm": 0.45838990807533264,
"learning_rate": 4.5454545454545455e-06,
"loss": 1.0069210529327393,
"step": 11
},
{
"epoch": 0.01643835616438356,
"grad_norm": 0.40803831815719604,
"learning_rate": 5e-06,
"loss": 1.0808029174804688,
"step": 12
},
{
"epoch": 0.01780821917808219,
"grad_norm": 0.40420499444007874,
"learning_rate": 5.4545454545454545e-06,
"loss": 1.0166305303573608,
"step": 13
},
{
"epoch": 0.019178082191780823,
"grad_norm": 0.3624785244464874,
"learning_rate": 5.90909090909091e-06,
"loss": 0.9620136618614197,
"step": 14
},
{
"epoch": 0.02054794520547945,
"grad_norm": 0.3782968521118164,
"learning_rate": 6.363636363636364e-06,
"loss": 1.0190021991729736,
"step": 15
},
{
"epoch": 0.021917808219178082,
"grad_norm": 0.3190377652645111,
"learning_rate": 6.818181818181818e-06,
"loss": 0.8993116617202759,
"step": 16
},
{
"epoch": 0.023287671232876714,
"grad_norm": 0.3656272888183594,
"learning_rate": 7.272727272727273e-06,
"loss": 1.0050773620605469,
"step": 17
},
{
"epoch": 0.024657534246575342,
"grad_norm": 0.2802956998348236,
"learning_rate": 7.727272727272727e-06,
"loss": 0.9016516804695129,
"step": 18
},
{
"epoch": 0.026027397260273973,
"grad_norm": 0.26026973128318787,
"learning_rate": 8.181818181818183e-06,
"loss": 0.8167034387588501,
"step": 19
},
{
"epoch": 0.0273972602739726,
"grad_norm": 0.2704644203186035,
"learning_rate": 8.636363636363637e-06,
"loss": 0.9548209309577942,
"step": 20
},
{
"epoch": 0.028767123287671233,
"grad_norm": 0.22691063582897186,
"learning_rate": 9.090909090909091e-06,
"loss": 0.8790600299835205,
"step": 21
},
{
"epoch": 0.030136986301369864,
"grad_norm": 0.22116470336914062,
"learning_rate": 9.545454545454547e-06,
"loss": 0.8698333501815796,
"step": 22
},
{
"epoch": 0.031506849315068496,
"grad_norm": 0.21417191624641418,
"learning_rate": 1e-05,
"loss": 0.8363385200500488,
"step": 23
},
{
"epoch": 0.03287671232876712,
"grad_norm": 0.24161672592163086,
"learning_rate": 1.0454545454545455e-05,
"loss": 0.9342292547225952,
"step": 24
},
{
"epoch": 0.03424657534246575,
"grad_norm": 0.21060842275619507,
"learning_rate": 1.0909090909090909e-05,
"loss": 0.8269792795181274,
"step": 25
},
{
"epoch": 0.03561643835616438,
"grad_norm": 0.2038818746805191,
"learning_rate": 1.1363636363636366e-05,
"loss": 0.7519025802612305,
"step": 26
},
{
"epoch": 0.036986301369863014,
"grad_norm": 0.2230827510356903,
"learning_rate": 1.181818181818182e-05,
"loss": 0.8076636791229248,
"step": 27
},
{
"epoch": 0.038356164383561646,
"grad_norm": 0.18758194148540497,
"learning_rate": 1.2272727272727274e-05,
"loss": 0.7615480422973633,
"step": 28
},
{
"epoch": 0.03972602739726028,
"grad_norm": 0.195050448179245,
"learning_rate": 1.2727272727272728e-05,
"loss": 0.8048295378684998,
"step": 29
},
{
"epoch": 0.0410958904109589,
"grad_norm": 0.1838914453983307,
"learning_rate": 1.3181818181818183e-05,
"loss": 0.7350910902023315,
"step": 30
},
{
"epoch": 0.04246575342465753,
"grad_norm": 0.20437878370285034,
"learning_rate": 1.3636363636363637e-05,
"loss": 0.8430003523826599,
"step": 31
},
{
"epoch": 0.043835616438356165,
"grad_norm": 0.20855139195919037,
"learning_rate": 1.4090909090909092e-05,
"loss": 0.752582848072052,
"step": 32
},
{
"epoch": 0.045205479452054796,
"grad_norm": 0.21374471485614777,
"learning_rate": 1.4545454545454546e-05,
"loss": 0.7721849679946899,
"step": 33
},
{
"epoch": 0.04657534246575343,
"grad_norm": 0.19961094856262207,
"learning_rate": 1.5000000000000002e-05,
"loss": 0.7583755850791931,
"step": 34
},
{
"epoch": 0.04794520547945205,
"grad_norm": 0.17798861861228943,
"learning_rate": 1.5454545454545454e-05,
"loss": 0.7548345327377319,
"step": 35
},
{
"epoch": 0.049315068493150684,
"grad_norm": 0.19771678745746613,
"learning_rate": 1.590909090909091e-05,
"loss": 0.7899003028869629,
"step": 36
},
{
"epoch": 0.050684931506849315,
"grad_norm": 0.18502211570739746,
"learning_rate": 1.6363636363636366e-05,
"loss": 0.7351113557815552,
"step": 37
},
{
"epoch": 0.052054794520547946,
"grad_norm": 0.2032163441181183,
"learning_rate": 1.681818181818182e-05,
"loss": 0.809609055519104,
"step": 38
},
{
"epoch": 0.05342465753424658,
"grad_norm": 0.17767126858234406,
"learning_rate": 1.7272727272727274e-05,
"loss": 0.6977474689483643,
"step": 39
},
{
"epoch": 0.0547945205479452,
"grad_norm": 0.1885196417570114,
"learning_rate": 1.772727272727273e-05,
"loss": 0.7495503425598145,
"step": 40
},
{
"epoch": 0.056164383561643834,
"grad_norm": 0.19037273526191711,
"learning_rate": 1.8181818181818182e-05,
"loss": 0.7687135338783264,
"step": 41
},
{
"epoch": 0.057534246575342465,
"grad_norm": 0.20152418315410614,
"learning_rate": 1.8636363636363638e-05,
"loss": 0.7619070410728455,
"step": 42
},
{
"epoch": 0.0589041095890411,
"grad_norm": 0.17043539881706238,
"learning_rate": 1.9090909090909094e-05,
"loss": 0.7616786360740662,
"step": 43
},
{
"epoch": 0.06027397260273973,
"grad_norm": 0.193017840385437,
"learning_rate": 1.9545454545454546e-05,
"loss": 0.6300766468048096,
"step": 44
},
{
"epoch": 0.06164383561643835,
"grad_norm": 0.17710553109645844,
"learning_rate": 2e-05,
"loss": 0.7160718441009521,
"step": 45
},
{
"epoch": 0.06301369863013699,
"grad_norm": 0.17088574171066284,
"learning_rate": 1.9999975388217704e-05,
"loss": 0.6980305314064026,
"step": 46
},
{
"epoch": 0.06438356164383562,
"grad_norm": 0.17629273235797882,
"learning_rate": 1.9999901552991966e-05,
"loss": 0.7771427035331726,
"step": 47
},
{
"epoch": 0.06575342465753424,
"grad_norm": 0.18566235899925232,
"learning_rate": 1.9999778494686226e-05,
"loss": 0.6564210057258606,
"step": 48
},
{
"epoch": 0.06712328767123288,
"grad_norm": 0.15922220051288605,
"learning_rate": 1.999960621390622e-05,
"loss": 0.6578161716461182,
"step": 49
},
{
"epoch": 0.0684931506849315,
"grad_norm": 0.17897970974445343,
"learning_rate": 1.999938471149998e-05,
"loss": 0.7257359623908997,
"step": 50
},
{
"epoch": 0.06986301369863014,
"grad_norm": 0.17765460908412933,
"learning_rate": 1.999911398855782e-05,
"loss": 0.7241641879081726,
"step": 51
},
{
"epoch": 0.07123287671232877,
"grad_norm": 0.1779472678899765,
"learning_rate": 1.999879404641233e-05,
"loss": 0.696056604385376,
"step": 52
},
{
"epoch": 0.07260273972602739,
"grad_norm": 0.18799914419651031,
"learning_rate": 1.9998424886638382e-05,
"loss": 0.7142607569694519,
"step": 53
},
{
"epoch": 0.07397260273972603,
"grad_norm": 0.17217737436294556,
"learning_rate": 1.999800651105311e-05,
"loss": 0.7063297033309937,
"step": 54
},
{
"epoch": 0.07534246575342465,
"grad_norm": 0.18024404346942902,
"learning_rate": 1.999753892171591e-05,
"loss": 0.741089940071106,
"step": 55
},
{
"epoch": 0.07671232876712329,
"grad_norm": 0.16077029705047607,
"learning_rate": 1.9997022120928423e-05,
"loss": 0.7010071277618408,
"step": 56
},
{
"epoch": 0.07808219178082192,
"grad_norm": 0.1703995168209076,
"learning_rate": 1.999645611123453e-05,
"loss": 0.6677719354629517,
"step": 57
},
{
"epoch": 0.07945205479452055,
"grad_norm": 0.16977134346961975,
"learning_rate": 1.9995840895420323e-05,
"loss": 0.7329045534133911,
"step": 58
},
{
"epoch": 0.08082191780821918,
"grad_norm": 0.17480944097042084,
"learning_rate": 1.999517647651412e-05,
"loss": 0.7148158550262451,
"step": 59
},
{
"epoch": 0.0821917808219178,
"grad_norm": 0.1859695464372635,
"learning_rate": 1.9994462857786433e-05,
"loss": 0.6804654598236084,
"step": 60
},
{
"epoch": 0.08356164383561644,
"grad_norm": 0.1692071110010147,
"learning_rate": 1.9993700042749937e-05,
"loss": 0.6517329216003418,
"step": 61
},
{
"epoch": 0.08493150684931507,
"grad_norm": 0.16654057800769806,
"learning_rate": 1.9992888035159484e-05,
"loss": 0.6754597425460815,
"step": 62
},
{
"epoch": 0.0863013698630137,
"grad_norm": 0.16556744277477264,
"learning_rate": 1.999202683901207e-05,
"loss": 0.6628267765045166,
"step": 63
},
{
"epoch": 0.08767123287671233,
"grad_norm": 0.17565493285655975,
"learning_rate": 1.99911164585468e-05,
"loss": 0.5974063873291016,
"step": 64
},
{
"epoch": 0.08904109589041095,
"grad_norm": 0.17702734470367432,
"learning_rate": 1.99901568982449e-05,
"loss": 0.7042257189750671,
"step": 65
},
{
"epoch": 0.09041095890410959,
"grad_norm": 0.18799234926700592,
"learning_rate": 1.9989148162829663e-05,
"loss": 0.6061130166053772,
"step": 66
},
{
"epoch": 0.09178082191780822,
"grad_norm": 0.1799335926771164,
"learning_rate": 1.9988090257266442e-05,
"loss": 0.6730984449386597,
"step": 67
},
{
"epoch": 0.09315068493150686,
"grad_norm": 0.18967068195343018,
"learning_rate": 1.998698318676263e-05,
"loss": 0.7190222144126892,
"step": 68
},
{
"epoch": 0.09452054794520548,
"grad_norm": 0.18353059887886047,
"learning_rate": 1.998582695676762e-05,
"loss": 0.660584568977356,
"step": 69
},
{
"epoch": 0.0958904109589041,
"grad_norm": 0.1780116707086563,
"learning_rate": 1.998462157297279e-05,
"loss": 0.7072656154632568,
"step": 70
},
{
"epoch": 0.09726027397260274,
"grad_norm": 0.16367685794830322,
"learning_rate": 1.998336704131147e-05,
"loss": 0.6226505041122437,
"step": 71
},
{
"epoch": 0.09863013698630137,
"grad_norm": 0.18768785893917084,
"learning_rate": 1.9982063367958907e-05,
"loss": 0.6319016218185425,
"step": 72
},
{
"epoch": 0.1,
"grad_norm": 0.17008233070373535,
"learning_rate": 1.9980710559332248e-05,
"loss": 0.6461281776428223,
"step": 73
},
{
"epoch": 0.10136986301369863,
"grad_norm": 0.18165260553359985,
"learning_rate": 1.9979308622090503e-05,
"loss": 0.6717009544372559,
"step": 74
},
{
"epoch": 0.10273972602739725,
"grad_norm": 0.17221364378929138,
"learning_rate": 1.9977857563134503e-05,
"loss": 0.6939575672149658,
"step": 75
},
{
"epoch": 0.10410958904109589,
"grad_norm": 0.16935478150844574,
"learning_rate": 1.9976357389606878e-05,
"loss": 0.6323124170303345,
"step": 76
},
{
"epoch": 0.10547945205479452,
"grad_norm": 0.18027523159980774,
"learning_rate": 1.9974808108892017e-05,
"loss": 0.6536504626274109,
"step": 77
},
{
"epoch": 0.10684931506849316,
"grad_norm": 0.17837466299533844,
"learning_rate": 1.9973209728616032e-05,
"loss": 0.6764005422592163,
"step": 78
},
{
"epoch": 0.10821917808219178,
"grad_norm": 0.17923392355442047,
"learning_rate": 1.997156225664672e-05,
"loss": 0.64453125,
"step": 79
},
{
"epoch": 0.1095890410958904,
"grad_norm": 0.17504896223545074,
"learning_rate": 1.996986570109353e-05,
"loss": 0.7095593214035034,
"step": 80
},
{
"epoch": 0.11095890410958904,
"grad_norm": 0.16979750990867615,
"learning_rate": 1.9968120070307503e-05,
"loss": 0.7190498113632202,
"step": 81
},
{
"epoch": 0.11232876712328767,
"grad_norm": 0.19454807043075562,
"learning_rate": 1.9966325372881265e-05,
"loss": 0.7216591238975525,
"step": 82
},
{
"epoch": 0.1136986301369863,
"grad_norm": 0.1744404286146164,
"learning_rate": 1.9964481617648954e-05,
"loss": 0.6529302597045898,
"step": 83
},
{
"epoch": 0.11506849315068493,
"grad_norm": 0.17792807519435883,
"learning_rate": 1.9962588813686187e-05,
"loss": 0.6740435361862183,
"step": 84
},
{
"epoch": 0.11643835616438356,
"grad_norm": 0.18751150369644165,
"learning_rate": 1.9960646970310027e-05,
"loss": 0.713350236415863,
"step": 85
},
{
"epoch": 0.1178082191780822,
"grad_norm": 0.1781667172908783,
"learning_rate": 1.9958656097078912e-05,
"loss": 0.649537205696106,
"step": 86
},
{
"epoch": 0.11917808219178082,
"grad_norm": 0.171499565243721,
"learning_rate": 1.9956616203792636e-05,
"loss": 0.6647717952728271,
"step": 87
},
{
"epoch": 0.12054794520547946,
"grad_norm": 0.1776268035173416,
"learning_rate": 1.9954527300492277e-05,
"loss": 0.6820968389511108,
"step": 88
},
{
"epoch": 0.12191780821917808,
"grad_norm": 0.1685784012079239,
"learning_rate": 1.995238939746016e-05,
"loss": 0.6538463234901428,
"step": 89
},
{
"epoch": 0.1232876712328767,
"grad_norm": 0.1900128871202469,
"learning_rate": 1.9950202505219808e-05,
"loss": 0.6783770322799683,
"step": 90
},
{
"epoch": 0.12465753424657534,
"grad_norm": 0.17134107649326324,
"learning_rate": 1.9947966634535887e-05,
"loss": 0.6427039504051208,
"step": 91
},
{
"epoch": 0.12602739726027398,
"grad_norm": 0.17828185856342316,
"learning_rate": 1.9945681796414146e-05,
"loss": 0.6815697550773621,
"step": 92
},
{
"epoch": 0.1273972602739726,
"grad_norm": 0.18462392687797546,
"learning_rate": 1.9943348002101374e-05,
"loss": 0.6666452288627625,
"step": 93
},
{
"epoch": 0.12876712328767123,
"grad_norm": 0.17112930119037628,
"learning_rate": 1.9940965263085335e-05,
"loss": 0.6950005292892456,
"step": 94
},
{
"epoch": 0.13013698630136986,
"grad_norm": 0.16788211464881897,
"learning_rate": 1.9938533591094728e-05,
"loss": 0.6137875318527222,
"step": 95
},
{
"epoch": 0.13150684931506848,
"grad_norm": 0.1813577562570572,
"learning_rate": 1.99360529980991e-05,
"loss": 0.7134063243865967,
"step": 96
},
{
"epoch": 0.13287671232876713,
"grad_norm": 0.17547398805618286,
"learning_rate": 1.993352349630882e-05,
"loss": 0.6599196791648865,
"step": 97
},
{
"epoch": 0.13424657534246576,
"grad_norm": 0.16298982501029968,
"learning_rate": 1.9930945098174994e-05,
"loss": 0.5956114530563354,
"step": 98
},
{
"epoch": 0.13561643835616438,
"grad_norm": 0.1693538874387741,
"learning_rate": 1.9928317816389416e-05,
"loss": 0.6117294430732727,
"step": 99
},
{
"epoch": 0.136986301369863,
"grad_norm": 0.17330867052078247,
"learning_rate": 1.992564166388451e-05,
"loss": 0.6786526441574097,
"step": 100
},
{
"epoch": 0.13835616438356163,
"grad_norm": 0.17630691826343536,
"learning_rate": 1.992291665383325e-05,
"loss": 0.7339305877685547,
"step": 101
},
{
"epoch": 0.13972602739726028,
"grad_norm": 0.18277214467525482,
"learning_rate": 1.9920142799649098e-05,
"loss": 0.6647192239761353,
"step": 102
},
{
"epoch": 0.1410958904109589,
"grad_norm": 0.1676325649023056,
"learning_rate": 1.9917320114985964e-05,
"loss": 0.7023746967315674,
"step": 103
},
{
"epoch": 0.14246575342465753,
"grad_norm": 0.17900800704956055,
"learning_rate": 1.9914448613738107e-05,
"loss": 0.6510761976242065,
"step": 104
},
{
"epoch": 0.14383561643835616,
"grad_norm": 0.15803074836730957,
"learning_rate": 1.9911528310040073e-05,
"loss": 0.6323767900466919,
"step": 105
},
{
"epoch": 0.14520547945205478,
"grad_norm": 0.190176859498024,
"learning_rate": 1.9908559218266644e-05,
"loss": 0.7440241575241089,
"step": 106
},
{
"epoch": 0.14657534246575343,
"grad_norm": 0.1805161088705063,
"learning_rate": 1.9905541353032744e-05,
"loss": 0.6736125946044922,
"step": 107
},
{
"epoch": 0.14794520547945206,
"grad_norm": 0.16393084824085236,
"learning_rate": 1.9902474729193385e-05,
"loss": 0.639750063419342,
"step": 108
},
{
"epoch": 0.14931506849315068,
"grad_norm": 0.1559421420097351,
"learning_rate": 1.989935936184358e-05,
"loss": 0.6098452210426331,
"step": 109
},
{
"epoch": 0.1506849315068493,
"grad_norm": 0.1748615801334381,
"learning_rate": 1.989619526631828e-05,
"loss": 0.6362285614013672,
"step": 110
},
{
"epoch": 0.15205479452054796,
"grad_norm": 0.15956446528434753,
"learning_rate": 1.9892982458192286e-05,
"loss": 0.6010555028915405,
"step": 111
},
{
"epoch": 0.15342465753424658,
"grad_norm": 0.17703907191753387,
"learning_rate": 1.988972095328019e-05,
"loss": 0.6596381664276123,
"step": 112
},
{
"epoch": 0.1547945205479452,
"grad_norm": 0.17541560530662537,
"learning_rate": 1.9886410767636284e-05,
"loss": 0.6365323066711426,
"step": 113
},
{
"epoch": 0.15616438356164383,
"grad_norm": 0.1705840677022934,
"learning_rate": 1.9883051917554473e-05,
"loss": 0.6401112079620361,
"step": 114
},
{
"epoch": 0.15753424657534246,
"grad_norm": 0.18547984957695007,
"learning_rate": 1.987964441956822e-05,
"loss": 0.680147647857666,
"step": 115
},
{
"epoch": 0.1589041095890411,
"grad_norm": 0.157182514667511,
"learning_rate": 1.9876188290450445e-05,
"loss": 0.6003460884094238,
"step": 116
},
{
"epoch": 0.16027397260273973,
"grad_norm": 0.1699901968240738,
"learning_rate": 1.9872683547213446e-05,
"loss": 0.5792864561080933,
"step": 117
},
{
"epoch": 0.16164383561643836,
"grad_norm": 0.17358699440956116,
"learning_rate": 1.986913020710882e-05,
"loss": 0.6301059722900391,
"step": 118
},
{
"epoch": 0.16301369863013698,
"grad_norm": 0.18124189972877502,
"learning_rate": 1.9865528287627372e-05,
"loss": 0.6473714113235474,
"step": 119
},
{
"epoch": 0.1643835616438356,
"grad_norm": 0.1570880264043808,
"learning_rate": 1.9861877806499033e-05,
"loss": 0.6518967747688293,
"step": 120
},
{
"epoch": 0.16575342465753426,
"grad_norm": 0.16372060775756836,
"learning_rate": 1.9858178781692777e-05,
"loss": 0.5721158981323242,
"step": 121
},
{
"epoch": 0.16712328767123288,
"grad_norm": 0.1550905704498291,
"learning_rate": 1.9854431231416515e-05,
"loss": 0.6409134864807129,
"step": 122
},
{
"epoch": 0.1684931506849315,
"grad_norm": 0.1671498715877533,
"learning_rate": 1.9850635174117033e-05,
"loss": 0.6450335383415222,
"step": 123
},
{
"epoch": 0.16986301369863013,
"grad_norm": 0.1591179221868515,
"learning_rate": 1.9846790628479874e-05,
"loss": 0.6999216079711914,
"step": 124
},
{
"epoch": 0.17123287671232876,
"grad_norm": 0.17138256132602692,
"learning_rate": 1.984289761342926e-05,
"loss": 0.6327433586120605,
"step": 125
},
{
"epoch": 0.1726027397260274,
"grad_norm": 0.18257306516170502,
"learning_rate": 1.9838956148128004e-05,
"loss": 0.6688971519470215,
"step": 126
},
{
"epoch": 0.17397260273972603,
"grad_norm": 0.15964193642139435,
"learning_rate": 1.98349662519774e-05,
"loss": 0.6297944188117981,
"step": 127
},
{
"epoch": 0.17534246575342466,
"grad_norm": 0.17521503567695618,
"learning_rate": 1.983092794461714e-05,
"loss": 0.6212207674980164,
"step": 128
},
{
"epoch": 0.17671232876712328,
"grad_norm": 0.15988564491271973,
"learning_rate": 1.982684124592521e-05,
"loss": 0.5915682315826416,
"step": 129
},
{
"epoch": 0.1780821917808219,
"grad_norm": 0.16260051727294922,
"learning_rate": 1.9822706176017802e-05,
"loss": 0.6029642820358276,
"step": 130
},
{
"epoch": 0.17945205479452056,
"grad_norm": 0.16412433981895447,
"learning_rate": 1.9818522755249203e-05,
"loss": 0.6668821573257446,
"step": 131
},
{
"epoch": 0.18082191780821918,
"grad_norm": 0.16716210544109344,
"learning_rate": 1.9814291004211695e-05,
"loss": 0.5958414077758789,
"step": 132
},
{
"epoch": 0.1821917808219178,
"grad_norm": 0.17294858396053314,
"learning_rate": 1.981001094373548e-05,
"loss": 0.6482897996902466,
"step": 133
},
{
"epoch": 0.18356164383561643,
"grad_norm": 0.17001552879810333,
"learning_rate": 1.980568259488852e-05,
"loss": 0.7122180461883545,
"step": 134
},
{
"epoch": 0.18493150684931506,
"grad_norm": 0.16195915639400482,
"learning_rate": 1.980130597897651e-05,
"loss": 0.584160327911377,
"step": 135
},
{
"epoch": 0.1863013698630137,
"grad_norm": 0.17711922526359558,
"learning_rate": 1.9796881117542703e-05,
"loss": 0.620893657207489,
"step": 136
},
{
"epoch": 0.18767123287671234,
"grad_norm": 0.1668933480978012,
"learning_rate": 1.979240803236785e-05,
"loss": 0.6142734885215759,
"step": 137
},
{
"epoch": 0.18904109589041096,
"grad_norm": 0.16164347529411316,
"learning_rate": 1.9787886745470067e-05,
"loss": 0.6293851137161255,
"step": 138
},
{
"epoch": 0.19041095890410958,
"grad_norm": 0.15106764435768127,
"learning_rate": 1.9783317279104736e-05,
"loss": 0.6398227214813232,
"step": 139
},
{
"epoch": 0.1917808219178082,
"grad_norm": 0.16852658987045288,
"learning_rate": 1.977869965576441e-05,
"loss": 0.6542407274246216,
"step": 140
},
{
"epoch": 0.19315068493150686,
"grad_norm": 0.16625036299228668,
"learning_rate": 1.9774033898178668e-05,
"loss": 0.6504607200622559,
"step": 141
},
{
"epoch": 0.19452054794520549,
"grad_norm": 0.16894933581352234,
"learning_rate": 1.9769320029314037e-05,
"loss": 0.6176894903182983,
"step": 142
},
{
"epoch": 0.1958904109589041,
"grad_norm": 0.15120083093643188,
"learning_rate": 1.9764558072373855e-05,
"loss": 0.566133975982666,
"step": 143
},
{
"epoch": 0.19726027397260273,
"grad_norm": 0.1807292252779007,
"learning_rate": 1.9759748050798176e-05,
"loss": 0.6475929021835327,
"step": 144
},
{
"epoch": 0.19863013698630136,
"grad_norm": 0.18464243412017822,
"learning_rate": 1.975488998826364e-05,
"loss": 0.6212013959884644,
"step": 145
},
{
"epoch": 0.2,
"grad_norm": 0.18034766614437103,
"learning_rate": 1.9749983908683362e-05,
"loss": 0.6008216142654419,
"step": 146
},
{
"epoch": 0.20136986301369864,
"grad_norm": 0.1733216792345047,
"learning_rate": 1.9745029836206813e-05,
"loss": 0.654267430305481,
"step": 147
},
{
"epoch": 0.20273972602739726,
"grad_norm": 0.1807546615600586,
"learning_rate": 1.9740027795219706e-05,
"loss": 0.6405000686645508,
"step": 148
},
{
"epoch": 0.20410958904109588,
"grad_norm": 0.17537952959537506,
"learning_rate": 1.9734977810343868e-05,
"loss": 0.6626770496368408,
"step": 149
},
{
"epoch": 0.2054794520547945,
"grad_norm": 0.16498853266239166,
"learning_rate": 1.9729879906437124e-05,
"loss": 0.6191319227218628,
"step": 150
},
{
"epoch": 0.20684931506849316,
"grad_norm": 0.1608179807662964,
"learning_rate": 1.972473410859318e-05,
"loss": 0.6513334512710571,
"step": 151
},
{
"epoch": 0.20821917808219179,
"grad_norm": 0.16782471537590027,
"learning_rate": 1.9719540442141477e-05,
"loss": 0.6400517225265503,
"step": 152
},
{
"epoch": 0.2095890410958904,
"grad_norm": 0.16582812368869781,
"learning_rate": 1.97142989326471e-05,
"loss": 0.6530036926269531,
"step": 153
},
{
"epoch": 0.21095890410958903,
"grad_norm": 0.1592300981283188,
"learning_rate": 1.9709009605910628e-05,
"loss": 0.5819565653800964,
"step": 154
},
{
"epoch": 0.21232876712328766,
"grad_norm": 0.1471843123435974,
"learning_rate": 1.970367248796801e-05,
"loss": 0.6289366483688354,
"step": 155
},
{
"epoch": 0.2136986301369863,
"grad_norm": 0.15616214275360107,
"learning_rate": 1.969828760509044e-05,
"loss": 0.6071575880050659,
"step": 156
},
{
"epoch": 0.21506849315068494,
"grad_norm": 0.15616996586322784,
"learning_rate": 1.9692854983784235e-05,
"loss": 0.570655345916748,
"step": 157
},
{
"epoch": 0.21643835616438356,
"grad_norm": 0.16667386889457703,
"learning_rate": 1.9687374650790693e-05,
"loss": 0.6458644866943359,
"step": 158
},
{
"epoch": 0.21780821917808219,
"grad_norm": 0.15973800420761108,
"learning_rate": 1.9681846633085968e-05,
"loss": 0.6385851502418518,
"step": 159
},
{
"epoch": 0.2191780821917808,
"grad_norm": 0.15689517557621002,
"learning_rate": 1.9676270957880928e-05,
"loss": 0.6954727172851562,
"step": 160
},
{
"epoch": 0.22054794520547946,
"grad_norm": 0.16799283027648926,
"learning_rate": 1.9670647652621044e-05,
"loss": 0.6535817384719849,
"step": 161
},
{
"epoch": 0.2219178082191781,
"grad_norm": 0.1637307107448578,
"learning_rate": 1.966497674498622e-05,
"loss": 0.6275250315666199,
"step": 162
},
{
"epoch": 0.2232876712328767,
"grad_norm": 0.1446286141872406,
"learning_rate": 1.9659258262890683e-05,
"loss": 0.6088711023330688,
"step": 163
},
{
"epoch": 0.22465753424657534,
"grad_norm": 0.16091319918632507,
"learning_rate": 1.965349223448285e-05,
"loss": 0.5717509984970093,
"step": 164
},
{
"epoch": 0.22602739726027396,
"grad_norm": 0.14229550957679749,
"learning_rate": 1.9647678688145163e-05,
"loss": 0.5792914628982544,
"step": 165
},
{
"epoch": 0.2273972602739726,
"grad_norm": 0.14515039324760437,
"learning_rate": 1.9641817652493964e-05,
"loss": 0.5595550537109375,
"step": 166
},
{
"epoch": 0.22876712328767124,
"grad_norm": 0.1727345734834671,
"learning_rate": 1.9635909156379373e-05,
"loss": 0.665512204170227,
"step": 167
},
{
"epoch": 0.23013698630136986,
"grad_norm": 0.14354562759399414,
"learning_rate": 1.96299532288851e-05,
"loss": 0.5523260235786438,
"step": 168
},
{
"epoch": 0.23150684931506849,
"grad_norm": 0.1551981419324875,
"learning_rate": 1.9623949899328352e-05,
"loss": 0.6456075310707092,
"step": 169
},
{
"epoch": 0.2328767123287671,
"grad_norm": 0.1689499169588089,
"learning_rate": 1.961789919725965e-05,
"loss": 0.621225118637085,
"step": 170
},
{
"epoch": 0.23424657534246576,
"grad_norm": 0.16959482431411743,
"learning_rate": 1.9611801152462715e-05,
"loss": 0.649752140045166,
"step": 171
},
{
"epoch": 0.2356164383561644,
"grad_norm": 0.1623905897140503,
"learning_rate": 1.960565579495429e-05,
"loss": 0.6222062110900879,
"step": 172
},
{
"epoch": 0.236986301369863,
"grad_norm": 0.16222091019153595,
"learning_rate": 1.959946315498402e-05,
"loss": 0.6512526869773865,
"step": 173
},
{
"epoch": 0.23835616438356164,
"grad_norm": 0.16222915053367615,
"learning_rate": 1.9593223263034288e-05,
"loss": 0.5998870134353638,
"step": 174
},
{
"epoch": 0.23972602739726026,
"grad_norm": 0.16737569868564606,
"learning_rate": 1.9586936149820056e-05,
"loss": 0.6641021966934204,
"step": 175
},
{
"epoch": 0.2410958904109589,
"grad_norm": 0.16463534533977509,
"learning_rate": 1.9580601846288747e-05,
"loss": 0.5590486526489258,
"step": 176
},
{
"epoch": 0.24246575342465754,
"grad_norm": 0.14511965215206146,
"learning_rate": 1.9574220383620054e-05,
"loss": 0.5980298519134521,
"step": 177
},
{
"epoch": 0.24383561643835616,
"grad_norm": 0.17374759912490845,
"learning_rate": 1.9567791793225818e-05,
"loss": 0.6144176721572876,
"step": 178
},
{
"epoch": 0.2452054794520548,
"grad_norm": 0.1671595722436905,
"learning_rate": 1.9561316106749847e-05,
"loss": 0.6108953952789307,
"step": 179
},
{
"epoch": 0.2465753424657534,
"grad_norm": 0.1550654023885727,
"learning_rate": 1.955479335606778e-05,
"loss": 0.563538670539856,
"step": 180
},
{
"epoch": 0.24794520547945206,
"grad_norm": 0.15254142880439758,
"learning_rate": 1.954822357328692e-05,
"loss": 0.605830192565918,
"step": 181
},
{
"epoch": 0.2493150684931507,
"grad_norm": 0.17636969685554504,
"learning_rate": 1.9541606790746082e-05,
"loss": 0.649669885635376,
"step": 182
},
{
"epoch": 0.25068493150684934,
"grad_norm": 0.16506697237491608,
"learning_rate": 1.9534943041015425e-05,
"loss": 0.6465010643005371,
"step": 183
},
{
"epoch": 0.25205479452054796,
"grad_norm": 0.16332168877124786,
"learning_rate": 1.95282323568963e-05,
"loss": 0.6338891386985779,
"step": 184
},
{
"epoch": 0.2534246575342466,
"grad_norm": 0.15076348185539246,
"learning_rate": 1.9521474771421093e-05,
"loss": 0.6196539998054504,
"step": 185
},
{
"epoch": 0.2547945205479452,
"grad_norm": 0.15440316498279572,
"learning_rate": 1.9514670317853043e-05,
"loss": 0.5998337864875793,
"step": 186
},
{
"epoch": 0.25616438356164384,
"grad_norm": 0.16012319922447205,
"learning_rate": 1.9507819029686094e-05,
"loss": 0.6014530658721924,
"step": 187
},
{
"epoch": 0.25753424657534246,
"grad_norm": 0.15418842434883118,
"learning_rate": 1.9500920940644734e-05,
"loss": 0.6077374219894409,
"step": 188
},
{
"epoch": 0.2589041095890411,
"grad_norm": 0.16678105294704437,
"learning_rate": 1.9493976084683814e-05,
"loss": 0.6204032897949219,
"step": 189
},
{
"epoch": 0.2602739726027397,
"grad_norm": 0.15462566912174225,
"learning_rate": 1.948698449598839e-05,
"loss": 0.5938429236412048,
"step": 190
},
{
"epoch": 0.26164383561643834,
"grad_norm": 0.15948761999607086,
"learning_rate": 1.9479946208973554e-05,
"loss": 0.5870481729507446,
"step": 191
},
{
"epoch": 0.26301369863013696,
"grad_norm": 0.17368458211421967,
"learning_rate": 1.947286125828426e-05,
"loss": 0.6516101360321045,
"step": 192
},
{
"epoch": 0.26438356164383564,
"grad_norm": 0.16198550164699554,
"learning_rate": 1.946572967879517e-05,
"loss": 0.5894348621368408,
"step": 193
},
{
"epoch": 0.26575342465753427,
"grad_norm": 0.15812379121780396,
"learning_rate": 1.9458551505610452e-05,
"loss": 0.6530876755714417,
"step": 194
},
{
"epoch": 0.2671232876712329,
"grad_norm": 0.15484529733657837,
"learning_rate": 1.9451326774063636e-05,
"loss": 0.5291624069213867,
"step": 195
},
{
"epoch": 0.2684931506849315,
"grad_norm": 0.14930173754692078,
"learning_rate": 1.944405551971743e-05,
"loss": 0.5648593306541443,
"step": 196
},
{
"epoch": 0.26986301369863014,
"grad_norm": 0.16413788497447968,
"learning_rate": 1.9436737778363526e-05,
"loss": 0.5979588031768799,
"step": 197
},
{
"epoch": 0.27123287671232876,
"grad_norm": 0.16401752829551697,
"learning_rate": 1.9429373586022472e-05,
"loss": 0.6155828237533569,
"step": 198
},
{
"epoch": 0.2726027397260274,
"grad_norm": 0.15918703377246857,
"learning_rate": 1.9421962978943442e-05,
"loss": 0.6395516991615295,
"step": 199
},
{
"epoch": 0.273972602739726,
"grad_norm": 0.16323615610599518,
"learning_rate": 1.9414505993604083e-05,
"loss": 0.6528345346450806,
"step": 200
},
{
"epoch": 0.27534246575342464,
"grad_norm": 0.17287138104438782,
"learning_rate": 1.9407002666710334e-05,
"loss": 0.6492872834205627,
"step": 201
},
{
"epoch": 0.27671232876712326,
"grad_norm": 0.17330391705036163,
"learning_rate": 1.939945303519625e-05,
"loss": 0.6085129380226135,
"step": 202
},
{
"epoch": 0.27808219178082194,
"grad_norm": 0.1582794040441513,
"learning_rate": 1.93918571362238e-05,
"loss": 0.6202000379562378,
"step": 203
},
{
"epoch": 0.27945205479452057,
"grad_norm": 0.15793833136558533,
"learning_rate": 1.938421500718272e-05,
"loss": 0.5806069374084473,
"step": 204
},
{
"epoch": 0.2808219178082192,
"grad_norm": 0.16570699214935303,
"learning_rate": 1.937652668569028e-05,
"loss": 0.6103482246398926,
"step": 205
},
{
"epoch": 0.2821917808219178,
"grad_norm": 0.1538430154323578,
"learning_rate": 1.9368792209591145e-05,
"loss": 0.5779464244842529,
"step": 206
},
{
"epoch": 0.28356164383561644,
"grad_norm": 0.1592852622270584,
"learning_rate": 1.9361011616957165e-05,
"loss": 0.6240794658660889,
"step": 207
},
{
"epoch": 0.28493150684931506,
"grad_norm": 0.1604919135570526,
"learning_rate": 1.935318494608718e-05,
"loss": 0.5771029591560364,
"step": 208
},
{
"epoch": 0.2863013698630137,
"grad_norm": 0.1573483794927597,
"learning_rate": 1.934531223550687e-05,
"loss": 0.6847480535507202,
"step": 209
},
{
"epoch": 0.2876712328767123,
"grad_norm": 0.15778696537017822,
"learning_rate": 1.933739352396851e-05,
"loss": 0.6099205017089844,
"step": 210
},
{
"epoch": 0.28904109589041094,
"grad_norm": 0.15076380968093872,
"learning_rate": 1.932942885045083e-05,
"loss": 0.5610252618789673,
"step": 211
},
{
"epoch": 0.29041095890410956,
"grad_norm": 0.1657320261001587,
"learning_rate": 1.9321418254158782e-05,
"loss": 0.6706528067588806,
"step": 212
},
{
"epoch": 0.29178082191780824,
"grad_norm": 0.1694263070821762,
"learning_rate": 1.9313361774523387e-05,
"loss": 0.6531721949577332,
"step": 213
},
{
"epoch": 0.29315068493150687,
"grad_norm": 0.15693512558937073,
"learning_rate": 1.9305259451201502e-05,
"loss": 0.6321283578872681,
"step": 214
},
{
"epoch": 0.2945205479452055,
"grad_norm": 0.16327035427093506,
"learning_rate": 1.929711132407565e-05,
"loss": 0.6443174481391907,
"step": 215
},
{
"epoch": 0.2958904109589041,
"grad_norm": 0.16162200272083282,
"learning_rate": 1.9288917433253823e-05,
"loss": 0.6530073881149292,
"step": 216
},
{
"epoch": 0.29726027397260274,
"grad_norm": 0.15862113237380981,
"learning_rate": 1.9280677819069273e-05,
"loss": 0.5875598788261414,
"step": 217
},
{
"epoch": 0.29863013698630136,
"grad_norm": 0.15702694654464722,
"learning_rate": 1.927239252208031e-05,
"loss": 0.5955379605293274,
"step": 218
},
{
"epoch": 0.3,
"grad_norm": 0.15328216552734375,
"learning_rate": 1.9264061583070126e-05,
"loss": 0.663190484046936,
"step": 219
},
{
"epoch": 0.3013698630136986,
"grad_norm": 0.17017608880996704,
"learning_rate": 1.9255685043046572e-05,
"loss": 0.6066758036613464,
"step": 220
},
{
"epoch": 0.30273972602739724,
"grad_norm": 0.1607287973165512,
"learning_rate": 1.924726294324196e-05,
"loss": 0.5664497017860413,
"step": 221
},
{
"epoch": 0.3041095890410959,
"grad_norm": 0.1684304177761078,
"learning_rate": 1.9238795325112867e-05,
"loss": 0.6179485321044922,
"step": 222
},
{
"epoch": 0.30547945205479454,
"grad_norm": 0.1446060985326767,
"learning_rate": 1.9230282230339936e-05,
"loss": 0.581000030040741,
"step": 223
},
{
"epoch": 0.30684931506849317,
"grad_norm": 0.16912856698036194,
"learning_rate": 1.9221723700827646e-05,
"loss": 0.6572670936584473,
"step": 224
},
{
"epoch": 0.3082191780821918,
"grad_norm": 0.16073641180992126,
"learning_rate": 1.921311977870413e-05,
"loss": 0.5523173809051514,
"step": 225
},
{
"epoch": 0.3095890410958904,
"grad_norm": 0.16843587160110474,
"learning_rate": 1.9204470506320968e-05,
"loss": 0.6724870204925537,
"step": 226
},
{
"epoch": 0.31095890410958904,
"grad_norm": 0.16108570992946625,
"learning_rate": 1.9195775926252952e-05,
"loss": 0.6259427070617676,
"step": 227
},
{
"epoch": 0.31232876712328766,
"grad_norm": 0.15122473239898682,
"learning_rate": 1.9187036081297907e-05,
"loss": 0.5658705234527588,
"step": 228
},
{
"epoch": 0.3136986301369863,
"grad_norm": 0.1597052663564682,
"learning_rate": 1.917825101447647e-05,
"loss": 0.6100412607192993,
"step": 229
},
{
"epoch": 0.3150684931506849,
"grad_norm": 0.18265262246131897,
"learning_rate": 1.9169420769031863e-05,
"loss": 0.5868899822235107,
"step": 230
},
{
"epoch": 0.31643835616438354,
"grad_norm": 0.15701507031917572,
"learning_rate": 1.916054538842971e-05,
"loss": 0.6006147265434265,
"step": 231
},
{
"epoch": 0.3178082191780822,
"grad_norm": 0.15414513647556305,
"learning_rate": 1.9151624916357793e-05,
"loss": 0.6486912965774536,
"step": 232
},
{
"epoch": 0.31917808219178084,
"grad_norm": 0.15333938598632812,
"learning_rate": 1.9142659396725862e-05,
"loss": 0.6258531808853149,
"step": 233
},
{
"epoch": 0.32054794520547947,
"grad_norm": 0.14656788110733032,
"learning_rate": 1.913364887366539e-05,
"loss": 0.5718272924423218,
"step": 234
},
{
"epoch": 0.3219178082191781,
"grad_norm": 0.16382336616516113,
"learning_rate": 1.912459339152939e-05,
"loss": 0.5820138454437256,
"step": 235
},
{
"epoch": 0.3232876712328767,
"grad_norm": 0.1630294770002365,
"learning_rate": 1.9115492994892172e-05,
"loss": 0.600603461265564,
"step": 236
},
{
"epoch": 0.32465753424657534,
"grad_norm": 0.14892295002937317,
"learning_rate": 1.9106347728549134e-05,
"loss": 0.5658777952194214,
"step": 237
},
{
"epoch": 0.32602739726027397,
"grad_norm": 0.16635073721408844,
"learning_rate": 1.9097157637516532e-05,
"loss": 0.5740564465522766,
"step": 238
},
{
"epoch": 0.3273972602739726,
"grad_norm": 0.16513577103614807,
"learning_rate": 1.9087922767031276e-05,
"loss": 0.6348574757575989,
"step": 239
},
{
"epoch": 0.3287671232876712,
"grad_norm": 0.16342297196388245,
"learning_rate": 1.9078643162550686e-05,
"loss": 0.6322500109672546,
"step": 240
},
{
"epoch": 0.33013698630136984,
"grad_norm": 0.1535942405462265,
"learning_rate": 1.906931886975228e-05,
"loss": 0.53902268409729,
"step": 241
},
{
"epoch": 0.3315068493150685,
"grad_norm": 0.14455796778202057,
"learning_rate": 1.905994993453356e-05,
"loss": 0.5630102157592773,
"step": 242
},
{
"epoch": 0.33287671232876714,
"grad_norm": 0.1663145124912262,
"learning_rate": 1.905053640301176e-05,
"loss": 0.6214050054550171,
"step": 243
},
{
"epoch": 0.33424657534246577,
"grad_norm": 0.1636534482240677,
"learning_rate": 1.9041078321523637e-05,
"loss": 0.6704033613204956,
"step": 244
},
{
"epoch": 0.3356164383561644,
"grad_norm": 0.1620258390903473,
"learning_rate": 1.903157573662524e-05,
"loss": 0.6345779895782471,
"step": 245
},
{
"epoch": 0.336986301369863,
"grad_norm": 0.15679852664470673,
"learning_rate": 1.9022028695091678e-05,
"loss": 0.5873825550079346,
"step": 246
},
{
"epoch": 0.33835616438356164,
"grad_norm": 0.15328912436962128,
"learning_rate": 1.9012437243916895e-05,
"loss": 0.6029560565948486,
"step": 247
},
{
"epoch": 0.33972602739726027,
"grad_norm": 0.14538195729255676,
"learning_rate": 1.900280143031343e-05,
"loss": 0.5177454948425293,
"step": 248
},
{
"epoch": 0.3410958904109589,
"grad_norm": 0.16042332351207733,
"learning_rate": 1.8993121301712194e-05,
"loss": 0.6091316342353821,
"step": 249
},
{
"epoch": 0.3424657534246575,
"grad_norm": 0.1434759795665741,
"learning_rate": 1.8983396905762232e-05,
"loss": 0.5697845220565796,
"step": 250
},
{
"epoch": 0.34383561643835614,
"grad_norm": 0.14016889035701752,
"learning_rate": 1.8973628290330484e-05,
"loss": 0.5575758218765259,
"step": 251
},
{
"epoch": 0.3452054794520548,
"grad_norm": 0.14783437550067902,
"learning_rate": 1.896381550350156e-05,
"loss": 0.5810775756835938,
"step": 252
},
{
"epoch": 0.34657534246575344,
"grad_norm": 0.15758486092090607,
"learning_rate": 1.895395859357749e-05,
"loss": 0.613425076007843,
"step": 253
},
{
"epoch": 0.34794520547945207,
"grad_norm": 0.15336844325065613,
"learning_rate": 1.8944057609077506e-05,
"loss": 0.5932927131652832,
"step": 254
},
{
"epoch": 0.3493150684931507,
"grad_norm": 0.1474757194519043,
"learning_rate": 1.8934112598737777e-05,
"loss": 0.5725477337837219,
"step": 255
},
{
"epoch": 0.3506849315068493,
"grad_norm": 0.14386139810085297,
"learning_rate": 1.892412361151119e-05,
"loss": 0.6071690320968628,
"step": 256
},
{
"epoch": 0.35205479452054794,
"grad_norm": 0.17297492921352386,
"learning_rate": 1.8914090696567104e-05,
"loss": 0.6448274850845337,
"step": 257
},
{
"epoch": 0.35342465753424657,
"grad_norm": 0.1554916501045227,
"learning_rate": 1.89040139032911e-05,
"loss": 0.6005516052246094,
"step": 258
},
{
"epoch": 0.3547945205479452,
"grad_norm": 0.15703563392162323,
"learning_rate": 1.8893893281284744e-05,
"loss": 0.5192084312438965,
"step": 259
},
{
"epoch": 0.3561643835616438,
"grad_norm": 0.1578558385372162,
"learning_rate": 1.8883728880365347e-05,
"loss": 0.6114726066589355,
"step": 260
},
{
"epoch": 0.35753424657534244,
"grad_norm": 0.16833758354187012,
"learning_rate": 1.8873520750565716e-05,
"loss": 0.6224363446235657,
"step": 261
},
{
"epoch": 0.3589041095890411,
"grad_norm": 0.1642412692308426,
"learning_rate": 1.8863268942133905e-05,
"loss": 0.6578483581542969,
"step": 262
},
{
"epoch": 0.36027397260273974,
"grad_norm": 0.16325688362121582,
"learning_rate": 1.8852973505532963e-05,
"loss": 0.6258366107940674,
"step": 263
},
{
"epoch": 0.36164383561643837,
"grad_norm": 0.16593174636363983,
"learning_rate": 1.8842634491440704e-05,
"loss": 0.6312078237533569,
"step": 264
},
{
"epoch": 0.363013698630137,
"grad_norm": 0.1558452844619751,
"learning_rate": 1.8832251950749443e-05,
"loss": 0.5959346294403076,
"step": 265
},
{
"epoch": 0.3643835616438356,
"grad_norm": 0.16329628229141235,
"learning_rate": 1.882182593456574e-05,
"loss": 0.5636122226715088,
"step": 266
},
{
"epoch": 0.36575342465753424,
"grad_norm": 0.1444038301706314,
"learning_rate": 1.8811356494210166e-05,
"loss": 0.5212979316711426,
"step": 267
},
{
"epoch": 0.36712328767123287,
"grad_norm": 0.16624850034713745,
"learning_rate": 1.880084368121704e-05,
"loss": 0.6462916135787964,
"step": 268
},
{
"epoch": 0.3684931506849315,
"grad_norm": 0.15940053761005402,
"learning_rate": 1.8790287547334178e-05,
"loss": 0.6105808615684509,
"step": 269
},
{
"epoch": 0.3698630136986301,
"grad_norm": 0.15389667451381683,
"learning_rate": 1.8779688144522625e-05,
"loss": 0.5553721785545349,
"step": 270
},
{
"epoch": 0.37123287671232874,
"grad_norm": 0.15304426848888397,
"learning_rate": 1.876904552495642e-05,
"loss": 0.6068332195281982,
"step": 271
},
{
"epoch": 0.3726027397260274,
"grad_norm": 0.1442302167415619,
"learning_rate": 1.875835974102234e-05,
"loss": 0.5685116052627563,
"step": 272
},
{
"epoch": 0.37397260273972605,
"grad_norm": 0.14790469408035278,
"learning_rate": 1.874763084531961e-05,
"loss": 0.5844176411628723,
"step": 273
},
{
"epoch": 0.37534246575342467,
"grad_norm": 0.14410603046417236,
"learning_rate": 1.8736858890659696e-05,
"loss": 0.5909885168075562,
"step": 274
},
{
"epoch": 0.3767123287671233,
"grad_norm": 0.17475005984306335,
"learning_rate": 1.8726043930065985e-05,
"loss": 0.6046225428581238,
"step": 275
},
{
"epoch": 0.3780821917808219,
"grad_norm": 0.15059982240200043,
"learning_rate": 1.871518601677357e-05,
"loss": 0.5838688015937805,
"step": 276
},
{
"epoch": 0.37945205479452054,
"grad_norm": 0.17032840847969055,
"learning_rate": 1.8704285204228973e-05,
"loss": 0.6261906623840332,
"step": 277
},
{
"epoch": 0.38082191780821917,
"grad_norm": 0.15075044333934784,
"learning_rate": 1.8693341546089878e-05,
"loss": 0.6051650047302246,
"step": 278
},
{
"epoch": 0.3821917808219178,
"grad_norm": 0.1473180502653122,
"learning_rate": 1.8682355096224873e-05,
"loss": 0.5551210641860962,
"step": 279
},
{
"epoch": 0.3835616438356164,
"grad_norm": 0.15265880525112152,
"learning_rate": 1.8671325908713182e-05,
"loss": 0.570478081703186,
"step": 280
},
{
"epoch": 0.38493150684931504,
"grad_norm": 0.16518573462963104,
"learning_rate": 1.866025403784439e-05,
"loss": 0.6145535707473755,
"step": 281
},
{
"epoch": 0.3863013698630137,
"grad_norm": 0.15201589465141296,
"learning_rate": 1.8649139538118196e-05,
"loss": 0.5385153293609619,
"step": 282
},
{
"epoch": 0.38767123287671235,
"grad_norm": 0.14805659651756287,
"learning_rate": 1.863798246424413e-05,
"loss": 0.628183126449585,
"step": 283
},
{
"epoch": 0.38904109589041097,
"grad_norm": 0.14637944102287292,
"learning_rate": 1.862678287114129e-05,
"loss": 0.62421715259552,
"step": 284
},
{
"epoch": 0.3904109589041096,
"grad_norm": 0.1630602329969406,
"learning_rate": 1.8615540813938063e-05,
"loss": 0.6616168022155762,
"step": 285
},
{
"epoch": 0.3917808219178082,
"grad_norm": 0.1550370305776596,
"learning_rate": 1.860425634797186e-05,
"loss": 0.5864919424057007,
"step": 286
},
{
"epoch": 0.39315068493150684,
"grad_norm": 0.14578670263290405,
"learning_rate": 1.8592929528788844e-05,
"loss": 0.545616626739502,
"step": 287
},
{
"epoch": 0.39452054794520547,
"grad_norm": 0.14833520352840424,
"learning_rate": 1.8581560412143663e-05,
"loss": 0.5978792905807495,
"step": 288
},
{
"epoch": 0.3958904109589041,
"grad_norm": 0.1507352888584137,
"learning_rate": 1.8570149053999154e-05,
"loss": 0.579099178314209,
"step": 289
},
{
"epoch": 0.3972602739726027,
"grad_norm": 0.14446204900741577,
"learning_rate": 1.8558695510526092e-05,
"loss": 0.5924318432807922,
"step": 290
},
{
"epoch": 0.39863013698630134,
"grad_norm": 0.15075939893722534,
"learning_rate": 1.8547199838102904e-05,
"loss": 0.5611657500267029,
"step": 291
},
{
"epoch": 0.4,
"grad_norm": 0.15933777391910553,
"learning_rate": 1.8535662093315386e-05,
"loss": 0.5471000671386719,
"step": 292
},
{
"epoch": 0.40136986301369865,
"grad_norm": 0.15519365668296814,
"learning_rate": 1.852408233295643e-05,
"loss": 0.6255986094474792,
"step": 293
},
{
"epoch": 0.40273972602739727,
"grad_norm": 0.15151312947273254,
"learning_rate": 1.851246061402574e-05,
"loss": 0.5669071674346924,
"step": 294
},
{
"epoch": 0.4041095890410959,
"grad_norm": 0.16434764862060547,
"learning_rate": 1.8500796993729563e-05,
"loss": 0.5679120421409607,
"step": 295
},
{
"epoch": 0.4054794520547945,
"grad_norm": 0.164402574300766,
"learning_rate": 1.8489091529480398e-05,
"loss": 0.6179913282394409,
"step": 296
},
{
"epoch": 0.40684931506849314,
"grad_norm": 0.15723305940628052,
"learning_rate": 1.8477344278896708e-05,
"loss": 0.5922874212265015,
"step": 297
},
{
"epoch": 0.40821917808219177,
"grad_norm": 0.1540336310863495,
"learning_rate": 1.8465555299802655e-05,
"loss": 0.5662035346031189,
"step": 298
},
{
"epoch": 0.4095890410958904,
"grad_norm": 0.17283470928668976,
"learning_rate": 1.8453724650227788e-05,
"loss": 0.6761048436164856,
"step": 299
},
{
"epoch": 0.410958904109589,
"grad_norm": 0.16036930680274963,
"learning_rate": 1.8441852388406788e-05,
"loss": 0.5548433065414429,
"step": 300
},
{
"epoch": 0.4123287671232877,
"grad_norm": 0.14102788269519806,
"learning_rate": 1.8429938572779154e-05,
"loss": 0.5084259510040283,
"step": 301
},
{
"epoch": 0.4136986301369863,
"grad_norm": 0.15470127761363983,
"learning_rate": 1.8417983261988933e-05,
"loss": 0.6092421412467957,
"step": 302
},
{
"epoch": 0.41506849315068495,
"grad_norm": 0.1494375318288803,
"learning_rate": 1.840598651488443e-05,
"loss": 0.5430501699447632,
"step": 303
},
{
"epoch": 0.41643835616438357,
"grad_norm": 0.14817532896995544,
"learning_rate": 1.8393948390517915e-05,
"loss": 0.5703598260879517,
"step": 304
},
{
"epoch": 0.4178082191780822,
"grad_norm": 0.15214185416698456,
"learning_rate": 1.8381868948145317e-05,
"loss": 0.6218011379241943,
"step": 305
},
{
"epoch": 0.4191780821917808,
"grad_norm": 0.14840349555015564,
"learning_rate": 1.8369748247225965e-05,
"loss": 0.6339513659477234,
"step": 306
},
{
"epoch": 0.42054794520547945,
"grad_norm": 0.14049845933914185,
"learning_rate": 1.8357586347422266e-05,
"loss": 0.5735560059547424,
"step": 307
},
{
"epoch": 0.42191780821917807,
"grad_norm": 0.14952462911605835,
"learning_rate": 1.8345383308599425e-05,
"loss": 0.5752905607223511,
"step": 308
},
{
"epoch": 0.4232876712328767,
"grad_norm": 0.1464746594429016,
"learning_rate": 1.833313919082515e-05,
"loss": 0.5919328927993774,
"step": 309
},
{
"epoch": 0.4246575342465753,
"grad_norm": 0.16054148972034454,
"learning_rate": 1.8320854054369354e-05,
"loss": 0.6318283081054688,
"step": 310
},
{
"epoch": 0.426027397260274,
"grad_norm": 0.14838211238384247,
"learning_rate": 1.8308527959703864e-05,
"loss": 0.5462095737457275,
"step": 311
},
{
"epoch": 0.4273972602739726,
"grad_norm": 0.14827798306941986,
"learning_rate": 1.82961609675021e-05,
"loss": 0.587713897228241,
"step": 312
},
{
"epoch": 0.42876712328767125,
"grad_norm": 0.14198598265647888,
"learning_rate": 1.828375313863881e-05,
"loss": 0.5431427955627441,
"step": 313
},
{
"epoch": 0.4301369863013699,
"grad_norm": 0.15106132626533508,
"learning_rate": 1.8271304534189754e-05,
"loss": 0.5700148940086365,
"step": 314
},
{
"epoch": 0.4315068493150685,
"grad_norm": 0.14311951398849487,
"learning_rate": 1.8258815215431395e-05,
"loss": 0.5781939625740051,
"step": 315
},
{
"epoch": 0.4328767123287671,
"grad_norm": 0.15375271439552307,
"learning_rate": 1.8246285243840617e-05,
"loss": 0.5642305612564087,
"step": 316
},
{
"epoch": 0.43424657534246575,
"grad_norm": 0.1475384384393692,
"learning_rate": 1.8233714681094405e-05,
"loss": 0.5396713018417358,
"step": 317
},
{
"epoch": 0.43561643835616437,
"grad_norm": 0.14523953199386597,
"learning_rate": 1.8221103589069553e-05,
"loss": 0.6184444427490234,
"step": 318
},
{
"epoch": 0.436986301369863,
"grad_norm": 0.1605875939130783,
"learning_rate": 1.8208452029842342e-05,
"loss": 0.5515637397766113,
"step": 319
},
{
"epoch": 0.4383561643835616,
"grad_norm": 0.14436954259872437,
"learning_rate": 1.8195760065688266e-05,
"loss": 0.6023062467575073,
"step": 320
},
{
"epoch": 0.4397260273972603,
"grad_norm": 0.13907724618911743,
"learning_rate": 1.818302775908169e-05,
"loss": 0.6034039258956909,
"step": 321
},
{
"epoch": 0.4410958904109589,
"grad_norm": 0.15003663301467896,
"learning_rate": 1.8170255172695572e-05,
"loss": 0.608570396900177,
"step": 322
},
{
"epoch": 0.44246575342465755,
"grad_norm": 0.16264384984970093,
"learning_rate": 1.815744236940113e-05,
"loss": 0.6284328699111938,
"step": 323
},
{
"epoch": 0.4438356164383562,
"grad_norm": 0.15702727437019348,
"learning_rate": 1.814458941226755e-05,
"loss": 0.5362198352813721,
"step": 324
},
{
"epoch": 0.4452054794520548,
"grad_norm": 0.1596728265285492,
"learning_rate": 1.813169636456167e-05,
"loss": 0.575636088848114,
"step": 325
},
{
"epoch": 0.4465753424657534,
"grad_norm": 0.1574542075395584,
"learning_rate": 1.8118763289747666e-05,
"loss": 0.6510027647018433,
"step": 326
},
{
"epoch": 0.44794520547945205,
"grad_norm": 0.14724481105804443,
"learning_rate": 1.810579025148674e-05,
"loss": 0.5525646209716797,
"step": 327
},
{
"epoch": 0.44931506849315067,
"grad_norm": 0.1698918640613556,
"learning_rate": 1.8092777313636813e-05,
"loss": 0.660180926322937,
"step": 328
},
{
"epoch": 0.4506849315068493,
"grad_norm": 0.15697503089904785,
"learning_rate": 1.8079724540252204e-05,
"loss": 0.5749319791793823,
"step": 329
},
{
"epoch": 0.4520547945205479,
"grad_norm": 0.1613071709871292,
"learning_rate": 1.8066631995583318e-05,
"loss": 0.5470437407493591,
"step": 330
},
{
"epoch": 0.4534246575342466,
"grad_norm": 0.14156115055084229,
"learning_rate": 1.8053499744076323e-05,
"loss": 0.5292378664016724,
"step": 331
},
{
"epoch": 0.4547945205479452,
"grad_norm": 0.14653272926807404,
"learning_rate": 1.8040327850372847e-05,
"loss": 0.6091462969779968,
"step": 332
},
{
"epoch": 0.45616438356164385,
"grad_norm": 0.16075585782527924,
"learning_rate": 1.8027116379309637e-05,
"loss": 0.6253776550292969,
"step": 333
},
{
"epoch": 0.4575342465753425,
"grad_norm": 0.1504659652709961,
"learning_rate": 1.801386539591827e-05,
"loss": 0.5656317472457886,
"step": 334
},
{
"epoch": 0.4589041095890411,
"grad_norm": 0.14729438722133636,
"learning_rate": 1.8000574965424813e-05,
"loss": 0.5766185522079468,
"step": 335
},
{
"epoch": 0.4602739726027397,
"grad_norm": 0.1585109382867813,
"learning_rate": 1.7987245153249496e-05,
"loss": 0.556631326675415,
"step": 336
},
{
"epoch": 0.46164383561643835,
"grad_norm": 0.15586094558238983,
"learning_rate": 1.7973876025006407e-05,
"loss": 0.5819780826568604,
"step": 337
},
{
"epoch": 0.46301369863013697,
"grad_norm": 0.14767439663410187,
"learning_rate": 1.796046764650316e-05,
"loss": 0.5321399569511414,
"step": 338
},
{
"epoch": 0.4643835616438356,
"grad_norm": 0.15002456307411194,
"learning_rate": 1.7947020083740575e-05,
"loss": 0.5469136834144592,
"step": 339
},
{
"epoch": 0.4657534246575342,
"grad_norm": 0.14220520853996277,
"learning_rate": 1.7933533402912354e-05,
"loss": 0.6022108793258667,
"step": 340
},
{
"epoch": 0.4671232876712329,
"grad_norm": 0.1343364119529724,
"learning_rate": 1.7920007670404738e-05,
"loss": 0.5860722064971924,
"step": 341
},
{
"epoch": 0.4684931506849315,
"grad_norm": 0.1453264057636261,
"learning_rate": 1.7906442952796212e-05,
"loss": 0.5470801591873169,
"step": 342
},
{
"epoch": 0.46986301369863015,
"grad_norm": 0.15289293229579926,
"learning_rate": 1.7892839316857146e-05,
"loss": 0.5834886431694031,
"step": 343
},
{
"epoch": 0.4712328767123288,
"grad_norm": 0.15057221055030823,
"learning_rate": 1.787919682954949e-05,
"loss": 0.5459625720977783,
"step": 344
},
{
"epoch": 0.4726027397260274,
"grad_norm": 0.14852501451969147,
"learning_rate": 1.786551555802643e-05,
"loss": 0.57718825340271,
"step": 345
},
{
"epoch": 0.473972602739726,
"grad_norm": 0.14916345477104187,
"learning_rate": 1.785179556963205e-05,
"loss": 0.5912124514579773,
"step": 346
},
{
"epoch": 0.47534246575342465,
"grad_norm": 0.14343911409378052,
"learning_rate": 1.7838036931901033e-05,
"loss": 0.5384149551391602,
"step": 347
},
{
"epoch": 0.4767123287671233,
"grad_norm": 0.13747356832027435,
"learning_rate": 1.7824239712558303e-05,
"loss": 0.5556282997131348,
"step": 348
},
{
"epoch": 0.4780821917808219,
"grad_norm": 0.153511181473732,
"learning_rate": 1.781040397951868e-05,
"loss": 0.5881044864654541,
"step": 349
},
{
"epoch": 0.4794520547945205,
"grad_norm": 0.1565730720758438,
"learning_rate": 1.7796529800886583e-05,
"loss": 0.564147412776947,
"step": 350
},
{
"epoch": 0.4808219178082192,
"grad_norm": 0.14074577391147614,
"learning_rate": 1.778261724495566e-05,
"loss": 0.5963798761367798,
"step": 351
},
{
"epoch": 0.4821917808219178,
"grad_norm": 0.1508810669183731,
"learning_rate": 1.7768666380208474e-05,
"loss": 0.6377789378166199,
"step": 352
},
{
"epoch": 0.48356164383561645,
"grad_norm": 0.14624693989753723,
"learning_rate": 1.775467727531615e-05,
"loss": 0.6010425090789795,
"step": 353
},
{
"epoch": 0.4849315068493151,
"grad_norm": 0.1480599194765091,
"learning_rate": 1.774064999913805e-05,
"loss": 0.5669881701469421,
"step": 354
},
{
"epoch": 0.4863013698630137,
"grad_norm": 0.1533871442079544,
"learning_rate": 1.772658462072143e-05,
"loss": 0.6225383877754211,
"step": 355
},
{
"epoch": 0.4876712328767123,
"grad_norm": 0.16357164084911346,
"learning_rate": 1.7712481209301094e-05,
"loss": 0.6383495926856995,
"step": 356
},
{
"epoch": 0.48904109589041095,
"grad_norm": 0.136616513133049,
"learning_rate": 1.7698339834299064e-05,
"loss": 0.5601040124893188,
"step": 357
},
{
"epoch": 0.4904109589041096,
"grad_norm": 0.15099939703941345,
"learning_rate": 1.7684160565324222e-05,
"loss": 0.5624288320541382,
"step": 358
},
{
"epoch": 0.4917808219178082,
"grad_norm": 0.14297816157341003,
"learning_rate": 1.766994347217199e-05,
"loss": 0.6051282286643982,
"step": 359
},
{
"epoch": 0.4931506849315068,
"grad_norm": 0.13573159277439117,
"learning_rate": 1.765568862482397e-05,
"loss": 0.5469596982002258,
"step": 360
},
{
"epoch": 0.4945205479452055,
"grad_norm": 0.1435568928718567,
"learning_rate": 1.764139609344759e-05,
"loss": 0.5686632394790649,
"step": 361
},
{
"epoch": 0.4958904109589041,
"grad_norm": 0.1436147540807724,
"learning_rate": 1.7627065948395795e-05,
"loss": 0.5006036758422852,
"step": 362
},
{
"epoch": 0.49726027397260275,
"grad_norm": 0.16107958555221558,
"learning_rate": 1.7612698260206668e-05,
"loss": 0.6271241903305054,
"step": 363
},
{
"epoch": 0.4986301369863014,
"grad_norm": 0.15104009211063385,
"learning_rate": 1.7598293099603087e-05,
"loss": 0.5952401161193848,
"step": 364
},
{
"epoch": 0.5,
"grad_norm": 0.1506115049123764,
"learning_rate": 1.7583850537492386e-05,
"loss": 0.567657470703125,
"step": 365
},
{
"epoch": 0.5013698630136987,
"grad_norm": 0.16260883212089539,
"learning_rate": 1.7569370644966007e-05,
"loss": 0.6197588443756104,
"step": 366
},
{
"epoch": 0.5027397260273972,
"grad_norm": 0.16213645040988922,
"learning_rate": 1.7554853493299142e-05,
"loss": 0.6336584687232971,
"step": 367
},
{
"epoch": 0.5041095890410959,
"grad_norm": 0.15891368687152863,
"learning_rate": 1.7540299153950386e-05,
"loss": 0.6455602645874023,
"step": 368
},
{
"epoch": 0.5054794520547945,
"grad_norm": 0.14739622175693512,
"learning_rate": 1.7525707698561383e-05,
"loss": 0.5473576188087463,
"step": 369
},
{
"epoch": 0.5068493150684932,
"grad_norm": 0.1491318643093109,
"learning_rate": 1.7511079198956487e-05,
"loss": 0.5691012144088745,
"step": 370
},
{
"epoch": 0.5082191780821917,
"grad_norm": 0.15580914914608002,
"learning_rate": 1.7496413727142375e-05,
"loss": 0.5861622095108032,
"step": 371
},
{
"epoch": 0.5095890410958904,
"grad_norm": 0.1470382958650589,
"learning_rate": 1.7481711355307735e-05,
"loss": 0.6198167204856873,
"step": 372
},
{
"epoch": 0.510958904109589,
"grad_norm": 0.1570439487695694,
"learning_rate": 1.746697215582288e-05,
"loss": 0.5717829465866089,
"step": 373
},
{
"epoch": 0.5123287671232877,
"grad_norm": 0.16358670592308044,
"learning_rate": 1.74521962012394e-05,
"loss": 0.5932368040084839,
"step": 374
},
{
"epoch": 0.5136986301369864,
"grad_norm": 0.15532813966274261,
"learning_rate": 1.7437383564289816e-05,
"loss": 0.5695762038230896,
"step": 375
},
{
"epoch": 0.5150684931506849,
"grad_norm": 0.15465544164180756,
"learning_rate": 1.7422534317887204e-05,
"loss": 0.6023073196411133,
"step": 376
},
{
"epoch": 0.5164383561643836,
"grad_norm": 0.1464887410402298,
"learning_rate": 1.740764853512485e-05,
"loss": 0.5693724155426025,
"step": 377
},
{
"epoch": 0.5178082191780822,
"grad_norm": 0.1357395201921463,
"learning_rate": 1.739272628927588e-05,
"loss": 0.536882221698761,
"step": 378
},
{
"epoch": 0.5191780821917809,
"grad_norm": 0.15610602498054504,
"learning_rate": 1.7377767653792915e-05,
"loss": 0.6062382459640503,
"step": 379
},
{
"epoch": 0.5205479452054794,
"grad_norm": 0.1343219131231308,
"learning_rate": 1.7362772702307677e-05,
"loss": 0.5632308125495911,
"step": 380
},
{
"epoch": 0.5219178082191781,
"grad_norm": 0.15014639496803284,
"learning_rate": 1.7347741508630673e-05,
"loss": 0.5424991250038147,
"step": 381
},
{
"epoch": 0.5232876712328767,
"grad_norm": 0.13980846107006073,
"learning_rate": 1.7332674146750796e-05,
"loss": 0.5513417720794678,
"step": 382
},
{
"epoch": 0.5246575342465754,
"grad_norm": 0.1392579823732376,
"learning_rate": 1.731757069083497e-05,
"loss": 0.5692585110664368,
"step": 383
},
{
"epoch": 0.5260273972602739,
"grad_norm": 0.1480334848165512,
"learning_rate": 1.7302431215227782e-05,
"loss": 0.540658712387085,
"step": 384
},
{
"epoch": 0.5273972602739726,
"grad_norm": 0.15066419541835785,
"learning_rate": 1.728725579445114e-05,
"loss": 0.5327310562133789,
"step": 385
},
{
"epoch": 0.5287671232876713,
"grad_norm": 0.14655138552188873,
"learning_rate": 1.727204450320387e-05,
"loss": 0.5699806213378906,
"step": 386
},
{
"epoch": 0.5301369863013699,
"grad_norm": 0.13703569769859314,
"learning_rate": 1.725679741636136e-05,
"loss": 0.5863059759140015,
"step": 387
},
{
"epoch": 0.5315068493150685,
"grad_norm": 0.1514001339673996,
"learning_rate": 1.7241514608975224e-05,
"loss": 0.5537671446800232,
"step": 388
},
{
"epoch": 0.5328767123287671,
"grad_norm": 0.14395487308502197,
"learning_rate": 1.7226196156272873e-05,
"loss": 0.60512775182724,
"step": 389
},
{
"epoch": 0.5342465753424658,
"grad_norm": 0.143971249461174,
"learning_rate": 1.7210842133657197e-05,
"loss": 0.5473227500915527,
"step": 390
},
{
"epoch": 0.5356164383561643,
"grad_norm": 0.15576298534870148,
"learning_rate": 1.719545261670617e-05,
"loss": 0.601518452167511,
"step": 391
},
{
"epoch": 0.536986301369863,
"grad_norm": 0.14324787259101868,
"learning_rate": 1.718002768117248e-05,
"loss": 0.5695725679397583,
"step": 392
},
{
"epoch": 0.5383561643835616,
"grad_norm": 0.14859525859355927,
"learning_rate": 1.7164567402983153e-05,
"loss": 0.6340662240982056,
"step": 393
},
{
"epoch": 0.5397260273972603,
"grad_norm": 0.13690751791000366,
"learning_rate": 1.7149071858239195e-05,
"loss": 0.5283052325248718,
"step": 394
},
{
"epoch": 0.541095890410959,
"grad_norm": 0.1423366665840149,
"learning_rate": 1.7133541123215196e-05,
"loss": 0.5996649265289307,
"step": 395
},
{
"epoch": 0.5424657534246575,
"grad_norm": 0.15085336565971375,
"learning_rate": 1.7117975274358975e-05,
"loss": 0.5998474955558777,
"step": 396
},
{
"epoch": 0.5438356164383562,
"grad_norm": 0.15080514550209045,
"learning_rate": 1.7102374388291182e-05,
"loss": 0.5708898901939392,
"step": 397
},
{
"epoch": 0.5452054794520548,
"grad_norm": 0.131844162940979,
"learning_rate": 1.7086738541804947e-05,
"loss": 0.5706642866134644,
"step": 398
},
{
"epoch": 0.5465753424657535,
"grad_norm": 0.1624789983034134,
"learning_rate": 1.7071067811865477e-05,
"loss": 0.5648869872093201,
"step": 399
},
{
"epoch": 0.547945205479452,
"grad_norm": 0.14977680146694183,
"learning_rate": 1.7055362275609687e-05,
"loss": 0.5466816425323486,
"step": 400
},
{
"epoch": 0.5493150684931507,
"grad_norm": 0.12936653196811676,
"learning_rate": 1.7039622010345832e-05,
"loss": 0.5807684659957886,
"step": 401
},
{
"epoch": 0.5506849315068493,
"grad_norm": 0.15499675273895264,
"learning_rate": 1.70238470935531e-05,
"loss": 0.5646220445632935,
"step": 402
},
{
"epoch": 0.552054794520548,
"grad_norm": 0.15363413095474243,
"learning_rate": 1.700803760288126e-05,
"loss": 0.5537654161453247,
"step": 403
},
{
"epoch": 0.5534246575342465,
"grad_norm": 0.14746566116809845,
"learning_rate": 1.699219361615026e-05,
"loss": 0.5865470170974731,
"step": 404
},
{
"epoch": 0.5547945205479452,
"grad_norm": 0.14045830070972443,
"learning_rate": 1.6976315211349848e-05,
"loss": 0.5434695482254028,
"step": 405
},
{
"epoch": 0.5561643835616439,
"grad_norm": 0.15868021547794342,
"learning_rate": 1.6960402466639195e-05,
"loss": 0.5476789474487305,
"step": 406
},
{
"epoch": 0.5575342465753425,
"grad_norm": 0.1462065726518631,
"learning_rate": 1.6944455460346503e-05,
"loss": 0.5196582078933716,
"step": 407
},
{
"epoch": 0.5589041095890411,
"grad_norm": 0.14001627266407013,
"learning_rate": 1.692847427096862e-05,
"loss": 0.6024141311645508,
"step": 408
},
{
"epoch": 0.5602739726027397,
"grad_norm": 0.15260235965251923,
"learning_rate": 1.6912458977170655e-05,
"loss": 0.6341221332550049,
"step": 409
},
{
"epoch": 0.5616438356164384,
"grad_norm": 0.14107351005077362,
"learning_rate": 1.6896409657785598e-05,
"loss": 0.5811542272567749,
"step": 410
},
{
"epoch": 0.563013698630137,
"grad_norm": 0.1302739977836609,
"learning_rate": 1.6880326391813917e-05,
"loss": 0.544140100479126,
"step": 411
},
{
"epoch": 0.5643835616438356,
"grad_norm": 0.1420375555753708,
"learning_rate": 1.6864209258423177e-05,
"loss": 0.534217119216919,
"step": 412
},
{
"epoch": 0.5657534246575342,
"grad_norm": 0.14836066961288452,
"learning_rate": 1.6848058336947657e-05,
"loss": 0.6192389726638794,
"step": 413
},
{
"epoch": 0.5671232876712329,
"grad_norm": 0.16043499112129211,
"learning_rate": 1.683187370688795e-05,
"loss": 0.5258642435073853,
"step": 414
},
{
"epoch": 0.5684931506849316,
"grad_norm": 0.14773382246494293,
"learning_rate": 1.6815655447910574e-05,
"loss": 0.5696762800216675,
"step": 415
},
{
"epoch": 0.5698630136986301,
"grad_norm": 0.14542289078235626,
"learning_rate": 1.6799403639847582e-05,
"loss": 0.5968167185783386,
"step": 416
},
{
"epoch": 0.5712328767123288,
"grad_norm": 0.14496225118637085,
"learning_rate": 1.6783118362696162e-05,
"loss": 0.5987216830253601,
"step": 417
},
{
"epoch": 0.5726027397260274,
"grad_norm": 0.14567093551158905,
"learning_rate": 1.6766799696618257e-05,
"loss": 0.5913962125778198,
"step": 418
},
{
"epoch": 0.5739726027397261,
"grad_norm": 0.15489841997623444,
"learning_rate": 1.6750447721940157e-05,
"loss": 0.6150340437889099,
"step": 419
},
{
"epoch": 0.5753424657534246,
"grad_norm": 0.13718822598457336,
"learning_rate": 1.6734062519152113e-05,
"loss": 0.5184439420700073,
"step": 420
},
{
"epoch": 0.5767123287671233,
"grad_norm": 0.14923328161239624,
"learning_rate": 1.671764416890793e-05,
"loss": 0.6138299703598022,
"step": 421
},
{
"epoch": 0.5780821917808219,
"grad_norm": 0.16161896288394928,
"learning_rate": 1.670119275202458e-05,
"loss": 0.6450883150100708,
"step": 422
},
{
"epoch": 0.5794520547945206,
"grad_norm": 0.16076239943504333,
"learning_rate": 1.6684708349481808e-05,
"loss": 0.6587648391723633,
"step": 423
},
{
"epoch": 0.5808219178082191,
"grad_norm": 0.1475391983985901,
"learning_rate": 1.666819104242171e-05,
"loss": 0.5829423069953918,
"step": 424
},
{
"epoch": 0.5821917808219178,
"grad_norm": 0.159859299659729,
"learning_rate": 1.6651640912148362e-05,
"loss": 0.5809226632118225,
"step": 425
},
{
"epoch": 0.5835616438356165,
"grad_norm": 0.1477399319410324,
"learning_rate": 1.6635058040127408e-05,
"loss": 0.6071959733963013,
"step": 426
},
{
"epoch": 0.584931506849315,
"grad_norm": 0.1518445909023285,
"learning_rate": 1.661844250798565e-05,
"loss": 0.5572406649589539,
"step": 427
},
{
"epoch": 0.5863013698630137,
"grad_norm": 0.14316509664058685,
"learning_rate": 1.6601794397510666e-05,
"loss": 0.5702495574951172,
"step": 428
},
{
"epoch": 0.5876712328767123,
"grad_norm": 0.13820317387580872,
"learning_rate": 1.658511379065039e-05,
"loss": 0.6022301912307739,
"step": 429
},
{
"epoch": 0.589041095890411,
"grad_norm": 0.1447012573480606,
"learning_rate": 1.6568400769512708e-05,
"loss": 0.5637190341949463,
"step": 430
},
{
"epoch": 0.5904109589041096,
"grad_norm": 0.15339450538158417,
"learning_rate": 1.6551655416365072e-05,
"loss": 0.5606421232223511,
"step": 431
},
{
"epoch": 0.5917808219178082,
"grad_norm": 0.1346173882484436,
"learning_rate": 1.653487781363408e-05,
"loss": 0.5061419606208801,
"step": 432
},
{
"epoch": 0.5931506849315068,
"grad_norm": 0.14965689182281494,
"learning_rate": 1.6518068043905075e-05,
"loss": 0.5600637197494507,
"step": 433
},
{
"epoch": 0.5945205479452055,
"grad_norm": 0.13996529579162598,
"learning_rate": 1.650122618992173e-05,
"loss": 0.5642286539077759,
"step": 434
},
{
"epoch": 0.5958904109589042,
"grad_norm": 0.14598211646080017,
"learning_rate": 1.6484352334585654e-05,
"loss": 0.5623951554298401,
"step": 435
},
{
"epoch": 0.5972602739726027,
"grad_norm": 0.15535297989845276,
"learning_rate": 1.6467446560955982e-05,
"loss": 0.5812850594520569,
"step": 436
},
{
"epoch": 0.5986301369863014,
"grad_norm": 0.15179552137851715,
"learning_rate": 1.6450508952248957e-05,
"loss": 0.5285508632659912,
"step": 437
},
{
"epoch": 0.6,
"grad_norm": 0.13990969955921173,
"learning_rate": 1.6433539591837527e-05,
"loss": 0.4808741807937622,
"step": 438
},
{
"epoch": 0.6013698630136987,
"grad_norm": 0.13985474407672882,
"learning_rate": 1.641653856325093e-05,
"loss": 0.526069164276123,
"step": 439
},
{
"epoch": 0.6027397260273972,
"grad_norm": 0.14668938517570496,
"learning_rate": 1.639950595017429e-05,
"loss": 0.55939781665802,
"step": 440
},
{
"epoch": 0.6041095890410959,
"grad_norm": 0.1409137099981308,
"learning_rate": 1.6382441836448203e-05,
"loss": 0.5505657196044922,
"step": 441
},
{
"epoch": 0.6054794520547945,
"grad_norm": 0.13390517234802246,
"learning_rate": 1.6365346306068315e-05,
"loss": 0.5604054927825928,
"step": 442
},
{
"epoch": 0.6068493150684932,
"grad_norm": 0.1515313684940338,
"learning_rate": 1.6348219443184927e-05,
"loss": 0.600440263748169,
"step": 443
},
{
"epoch": 0.6082191780821918,
"grad_norm": 0.15308818221092224,
"learning_rate": 1.633106133210255e-05,
"loss": 0.5832364559173584,
"step": 444
},
{
"epoch": 0.6095890410958904,
"grad_norm": 0.162045419216156,
"learning_rate": 1.6313872057279536e-05,
"loss": 0.577210009098053,
"step": 445
},
{
"epoch": 0.6109589041095891,
"grad_norm": 0.14119122922420502,
"learning_rate": 1.629665170332762e-05,
"loss": 0.5397899746894836,
"step": 446
},
{
"epoch": 0.6123287671232877,
"grad_norm": 0.13274981081485748,
"learning_rate": 1.627940035501152e-05,
"loss": 0.4862290024757385,
"step": 447
},
{
"epoch": 0.6136986301369863,
"grad_norm": 0.14096492528915405,
"learning_rate": 1.6262118097248524e-05,
"loss": 0.5487799048423767,
"step": 448
},
{
"epoch": 0.6150684931506849,
"grad_norm": 0.12473436444997787,
"learning_rate": 1.624480501510806e-05,
"loss": 0.505416989326477,
"step": 449
},
{
"epoch": 0.6164383561643836,
"grad_norm": 0.16270937025547028,
"learning_rate": 1.62274611938113e-05,
"loss": 0.5941034555435181,
"step": 450
},
{
"epoch": 0.6178082191780822,
"grad_norm": 0.1519913524389267,
"learning_rate": 1.62100867187307e-05,
"loss": 0.589063286781311,
"step": 451
},
{
"epoch": 0.6191780821917808,
"grad_norm": 0.1484452784061432,
"learning_rate": 1.619268167538963e-05,
"loss": 0.6294279098510742,
"step": 452
},
{
"epoch": 0.6205479452054794,
"grad_norm": 0.1418556571006775,
"learning_rate": 1.617524614946192e-05,
"loss": 0.6004430055618286,
"step": 453
},
{
"epoch": 0.6219178082191781,
"grad_norm": 0.14398352801799774,
"learning_rate": 1.6157780226771437e-05,
"loss": 0.5689352750778198,
"step": 454
},
{
"epoch": 0.6232876712328768,
"grad_norm": 0.14114904403686523,
"learning_rate": 1.614028399329168e-05,
"loss": 0.571104884147644,
"step": 455
},
{
"epoch": 0.6246575342465753,
"grad_norm": 0.14283505082130432,
"learning_rate": 1.6122757535145346e-05,
"loss": 0.5134957432746887,
"step": 456
},
{
"epoch": 0.626027397260274,
"grad_norm": 0.14088678359985352,
"learning_rate": 1.6105200938603917e-05,
"loss": 0.6189490556716919,
"step": 457
},
{
"epoch": 0.6273972602739726,
"grad_norm": 0.1482560932636261,
"learning_rate": 1.608761429008721e-05,
"loss": 0.5793185234069824,
"step": 458
},
{
"epoch": 0.6287671232876713,
"grad_norm": 0.16769099235534668,
"learning_rate": 1.606999767616298e-05,
"loss": 0.5795351266860962,
"step": 459
},
{
"epoch": 0.6301369863013698,
"grad_norm": 0.14694984257221222,
"learning_rate": 1.6052351183546478e-05,
"loss": 0.6406443119049072,
"step": 460
},
{
"epoch": 0.6315068493150685,
"grad_norm": 0.1542738825082779,
"learning_rate": 1.603467489910004e-05,
"loss": 0.5569589138031006,
"step": 461
},
{
"epoch": 0.6328767123287671,
"grad_norm": 0.1484406739473343,
"learning_rate": 1.6016968909832632e-05,
"loss": 0.563373863697052,
"step": 462
},
{
"epoch": 0.6342465753424658,
"grad_norm": 0.14980265498161316,
"learning_rate": 1.5999233302899444e-05,
"loss": 0.5926274657249451,
"step": 463
},
{
"epoch": 0.6356164383561644,
"grad_norm": 0.14588281512260437,
"learning_rate": 1.5981468165601462e-05,
"loss": 0.5282426476478577,
"step": 464
},
{
"epoch": 0.636986301369863,
"grad_norm": 0.14488136768341064,
"learning_rate": 1.5963673585385016e-05,
"loss": 0.5637092590332031,
"step": 465
},
{
"epoch": 0.6383561643835617,
"grad_norm": 0.14770545065402985,
"learning_rate": 1.594584964984138e-05,
"loss": 0.5751156806945801,
"step": 466
},
{
"epoch": 0.6397260273972603,
"grad_norm": 0.14717961847782135,
"learning_rate": 1.5927996446706308e-05,
"loss": 0.5032989382743835,
"step": 467
},
{
"epoch": 0.6410958904109589,
"grad_norm": 0.1413588970899582,
"learning_rate": 1.591011406385964e-05,
"loss": 0.5489969253540039,
"step": 468
},
{
"epoch": 0.6424657534246575,
"grad_norm": 0.13199076056480408,
"learning_rate": 1.5892202589324835e-05,
"loss": 0.5251445770263672,
"step": 469
},
{
"epoch": 0.6438356164383562,
"grad_norm": 0.14698363840579987,
"learning_rate": 1.587426211126856e-05,
"loss": 0.5794821977615356,
"step": 470
},
{
"epoch": 0.6452054794520548,
"grad_norm": 0.139966681599617,
"learning_rate": 1.5856292718000235e-05,
"loss": 0.5483118891716003,
"step": 471
},
{
"epoch": 0.6465753424657534,
"grad_norm": 0.1384710967540741,
"learning_rate": 1.5838294497971622e-05,
"loss": 0.5625190734863281,
"step": 472
},
{
"epoch": 0.647945205479452,
"grad_norm": 0.15267488360404968,
"learning_rate": 1.582026753977638e-05,
"loss": 0.5484758615493774,
"step": 473
},
{
"epoch": 0.6493150684931507,
"grad_norm": 0.15189437568187714,
"learning_rate": 1.5802211932149614e-05,
"loss": 0.5427678823471069,
"step": 474
},
{
"epoch": 0.6506849315068494,
"grad_norm": 0.1514596790075302,
"learning_rate": 1.5784127763967473e-05,
"loss": 0.5608749389648438,
"step": 475
},
{
"epoch": 0.6520547945205479,
"grad_norm": 0.13448850810527802,
"learning_rate": 1.5766015124246664e-05,
"loss": 0.551884114742279,
"step": 476
},
{
"epoch": 0.6534246575342466,
"grad_norm": 0.13851968944072723,
"learning_rate": 1.5747874102144073e-05,
"loss": 0.5595521330833435,
"step": 477
},
{
"epoch": 0.6547945205479452,
"grad_norm": 0.14666830003261566,
"learning_rate": 1.5729704786956262e-05,
"loss": 0.5839216113090515,
"step": 478
},
{
"epoch": 0.6561643835616439,
"grad_norm": 0.15627364814281464,
"learning_rate": 1.5711507268119087e-05,
"loss": 0.5988858938217163,
"step": 479
},
{
"epoch": 0.6575342465753424,
"grad_norm": 0.1463092565536499,
"learning_rate": 1.5693281635207214e-05,
"loss": 0.537287175655365,
"step": 480
},
{
"epoch": 0.6589041095890411,
"grad_norm": 0.1374531239271164,
"learning_rate": 1.567502797793372e-05,
"loss": 0.5598975419998169,
"step": 481
},
{
"epoch": 0.6602739726027397,
"grad_norm": 0.12984918057918549,
"learning_rate": 1.5656746386149596e-05,
"loss": 0.5133352279663086,
"step": 482
},
{
"epoch": 0.6616438356164384,
"grad_norm": 0.15310728549957275,
"learning_rate": 1.563843694984336e-05,
"loss": 0.570267915725708,
"step": 483
},
{
"epoch": 0.663013698630137,
"grad_norm": 0.16105115413665771,
"learning_rate": 1.5620099759140587e-05,
"loss": 0.6069197654724121,
"step": 484
},
{
"epoch": 0.6643835616438356,
"grad_norm": 0.14937542378902435,
"learning_rate": 1.560173490430346e-05,
"loss": 0.582807719707489,
"step": 485
},
{
"epoch": 0.6657534246575343,
"grad_norm": 0.14819355309009552,
"learning_rate": 1.558334247573035e-05,
"loss": 0.4967902898788452,
"step": 486
},
{
"epoch": 0.6671232876712329,
"grad_norm": 0.13898441195487976,
"learning_rate": 1.5564922563955337e-05,
"loss": 0.5023531317710876,
"step": 487
},
{
"epoch": 0.6684931506849315,
"grad_norm": 0.1410035490989685,
"learning_rate": 1.5546475259647795e-05,
"loss": 0.5132713317871094,
"step": 488
},
{
"epoch": 0.6698630136986301,
"grad_norm": 0.1309722661972046,
"learning_rate": 1.5528000653611935e-05,
"loss": 0.5644882321357727,
"step": 489
},
{
"epoch": 0.6712328767123288,
"grad_norm": 0.13525459170341492,
"learning_rate": 1.550949883678635e-05,
"loss": 0.5111340284347534,
"step": 490
},
{
"epoch": 0.6726027397260274,
"grad_norm": 0.14471714198589325,
"learning_rate": 1.549096990024358e-05,
"loss": 0.5402161478996277,
"step": 491
},
{
"epoch": 0.673972602739726,
"grad_norm": 0.1380838304758072,
"learning_rate": 1.5472413935189656e-05,
"loss": 0.57719486951828,
"step": 492
},
{
"epoch": 0.6753424657534246,
"grad_norm": 0.1449180692434311,
"learning_rate": 1.5453831032963648e-05,
"loss": 0.4582573175430298,
"step": 493
},
{
"epoch": 0.6767123287671233,
"grad_norm": 0.16058871150016785,
"learning_rate": 1.543522128503723e-05,
"loss": 0.6209670305252075,
"step": 494
},
{
"epoch": 0.678082191780822,
"grad_norm": 0.1415194571018219,
"learning_rate": 1.541658478301421e-05,
"loss": 0.542889416217804,
"step": 495
},
{
"epoch": 0.6794520547945205,
"grad_norm": 0.13594785332679749,
"learning_rate": 1.53979216186301e-05,
"loss": 0.5576387047767639,
"step": 496
},
{
"epoch": 0.6808219178082192,
"grad_norm": 0.16646577417850494,
"learning_rate": 1.537923188375164e-05,
"loss": 0.6171765327453613,
"step": 497
},
{
"epoch": 0.6821917808219178,
"grad_norm": 0.14226126670837402,
"learning_rate": 1.5360515670376373e-05,
"loss": 0.5574626922607422,
"step": 498
},
{
"epoch": 0.6835616438356165,
"grad_norm": 0.14714594185352325,
"learning_rate": 1.5341773070632172e-05,
"loss": 0.5930618643760681,
"step": 499
},
{
"epoch": 0.684931506849315,
"grad_norm": 0.12718486785888672,
"learning_rate": 1.5323004176776796e-05,
"loss": 0.49652647972106934,
"step": 500
},
{
"epoch": 0.6863013698630137,
"grad_norm": 0.13479647040367126,
"learning_rate": 1.5304209081197425e-05,
"loss": 0.5470659732818604,
"step": 501
},
{
"epoch": 0.6876712328767123,
"grad_norm": 0.1462414562702179,
"learning_rate": 1.5285387876410224e-05,
"loss": 0.558161735534668,
"step": 502
},
{
"epoch": 0.689041095890411,
"grad_norm": 0.1436636745929718,
"learning_rate": 1.5266540655059872e-05,
"loss": 0.5366897583007812,
"step": 503
},
{
"epoch": 0.6904109589041096,
"grad_norm": 0.15740245580673218,
"learning_rate": 1.5247667509919104e-05,
"loss": 0.5611869096755981,
"step": 504
},
{
"epoch": 0.6917808219178082,
"grad_norm": 0.1479470580816269,
"learning_rate": 1.5228768533888278e-05,
"loss": 0.5563034415245056,
"step": 505
},
{
"epoch": 0.6931506849315069,
"grad_norm": 0.14391250908374786,
"learning_rate": 1.520984381999488e-05,
"loss": 0.5359901189804077,
"step": 506
},
{
"epoch": 0.6945205479452055,
"grad_norm": 0.14507880806922913,
"learning_rate": 1.5190893461393108e-05,
"loss": 0.5985116958618164,
"step": 507
},
{
"epoch": 0.6958904109589041,
"grad_norm": 0.12985731661319733,
"learning_rate": 1.5171917551363376e-05,
"loss": 0.5191558003425598,
"step": 508
},
{
"epoch": 0.6972602739726027,
"grad_norm": 0.13191670179367065,
"learning_rate": 1.5152916183311881e-05,
"loss": 0.5371098518371582,
"step": 509
},
{
"epoch": 0.6986301369863014,
"grad_norm": 0.15381883084774017,
"learning_rate": 1.5133889450770122e-05,
"loss": 0.582601010799408,
"step": 510
},
{
"epoch": 0.7,
"grad_norm": 0.13812492787837982,
"learning_rate": 1.5114837447394469e-05,
"loss": 0.5940638780593872,
"step": 511
},
{
"epoch": 0.7013698630136986,
"grad_norm": 0.14924225211143494,
"learning_rate": 1.5095760266965668e-05,
"loss": 0.5940001010894775,
"step": 512
},
{
"epoch": 0.7027397260273973,
"grad_norm": 0.14933045208454132,
"learning_rate": 1.50766580033884e-05,
"loss": 0.62128746509552,
"step": 513
},
{
"epoch": 0.7041095890410959,
"grad_norm": 0.15632203221321106,
"learning_rate": 1.5057530750690822e-05,
"loss": 0.5551106929779053,
"step": 514
},
{
"epoch": 0.7054794520547946,
"grad_norm": 0.13439548015594482,
"learning_rate": 1.5038378603024085e-05,
"loss": 0.5119856595993042,
"step": 515
},
{
"epoch": 0.7068493150684931,
"grad_norm": 0.14251737296581268,
"learning_rate": 1.5019201654661886e-05,
"loss": 0.5376251935958862,
"step": 516
},
{
"epoch": 0.7082191780821918,
"grad_norm": 0.1485617607831955,
"learning_rate": 1.5000000000000002e-05,
"loss": 0.5784671306610107,
"step": 517
},
{
"epoch": 0.7095890410958904,
"grad_norm": 0.1372535526752472,
"learning_rate": 1.4980773733555823e-05,
"loss": 0.558871865272522,
"step": 518
},
{
"epoch": 0.7109589041095891,
"grad_norm": 0.1411508321762085,
"learning_rate": 1.4961522949967887e-05,
"loss": 0.5023566484451294,
"step": 519
},
{
"epoch": 0.7123287671232876,
"grad_norm": 0.15227437019348145,
"learning_rate": 1.4942247743995407e-05,
"loss": 0.5967291593551636,
"step": 520
},
{
"epoch": 0.7136986301369863,
"grad_norm": 0.15094850957393646,
"learning_rate": 1.4922948210517822e-05,
"loss": 0.6140634417533875,
"step": 521
},
{
"epoch": 0.7150684931506849,
"grad_norm": 0.1562165468931198,
"learning_rate": 1.4903624444534317e-05,
"loss": 0.5756810307502747,
"step": 522
},
{
"epoch": 0.7164383561643836,
"grad_norm": 0.14749285578727722,
"learning_rate": 1.4884276541163351e-05,
"loss": 0.5795880556106567,
"step": 523
},
{
"epoch": 0.7178082191780822,
"grad_norm": 0.13962800800800323,
"learning_rate": 1.4864904595642206e-05,
"loss": 0.5142415761947632,
"step": 524
},
{
"epoch": 0.7191780821917808,
"grad_norm": 0.14828866720199585,
"learning_rate": 1.4845508703326504e-05,
"loss": 0.5461621284484863,
"step": 525
},
{
"epoch": 0.7205479452054795,
"grad_norm": 0.14395621418952942,
"learning_rate": 1.4826088959689734e-05,
"loss": 0.6380258202552795,
"step": 526
},
{
"epoch": 0.7219178082191781,
"grad_norm": 0.15022049844264984,
"learning_rate": 1.4806645460322804e-05,
"loss": 0.5415664315223694,
"step": 527
},
{
"epoch": 0.7232876712328767,
"grad_norm": 0.14218366146087646,
"learning_rate": 1.4787178300933543e-05,
"loss": 0.5103275775909424,
"step": 528
},
{
"epoch": 0.7246575342465753,
"grad_norm": 0.15063804388046265,
"learning_rate": 1.4767687577346253e-05,
"loss": 0.5823582410812378,
"step": 529
},
{
"epoch": 0.726027397260274,
"grad_norm": 0.14951996505260468,
"learning_rate": 1.4748173385501222e-05,
"loss": 0.5916295051574707,
"step": 530
},
{
"epoch": 0.7273972602739726,
"grad_norm": 0.15501154959201813,
"learning_rate": 1.4728635821454255e-05,
"loss": 0.5518443584442139,
"step": 531
},
{
"epoch": 0.7287671232876712,
"grad_norm": 0.15150362253189087,
"learning_rate": 1.470907498137621e-05,
"loss": 0.5295286178588867,
"step": 532
},
{
"epoch": 0.7301369863013699,
"grad_norm": 0.14389854669570923,
"learning_rate": 1.4689490961552513e-05,
"loss": 0.573415994644165,
"step": 533
},
{
"epoch": 0.7315068493150685,
"grad_norm": 0.14579284191131592,
"learning_rate": 1.4669883858382689e-05,
"loss": 0.5175173282623291,
"step": 534
},
{
"epoch": 0.7328767123287672,
"grad_norm": 0.16025300323963165,
"learning_rate": 1.4650253768379896e-05,
"loss": 0.5780712366104126,
"step": 535
},
{
"epoch": 0.7342465753424657,
"grad_norm": 0.15502531826496124,
"learning_rate": 1.4630600788170429e-05,
"loss": 0.5766080617904663,
"step": 536
},
{
"epoch": 0.7356164383561644,
"grad_norm": 0.1452580839395523,
"learning_rate": 1.461092501449326e-05,
"loss": 0.5627033710479736,
"step": 537
},
{
"epoch": 0.736986301369863,
"grad_norm": 0.14218607544898987,
"learning_rate": 1.4591226544199564e-05,
"loss": 0.5479629635810852,
"step": 538
},
{
"epoch": 0.7383561643835617,
"grad_norm": 0.14063885807991028,
"learning_rate": 1.4571505474252234e-05,
"loss": 0.5578610301017761,
"step": 539
},
{
"epoch": 0.7397260273972602,
"grad_norm": 0.14620015025138855,
"learning_rate": 1.4551761901725402e-05,
"loss": 0.5858629941940308,
"step": 540
},
{
"epoch": 0.7410958904109589,
"grad_norm": 0.14063887298107147,
"learning_rate": 1.4531995923803974e-05,
"loss": 0.5905371904373169,
"step": 541
},
{
"epoch": 0.7424657534246575,
"grad_norm": 0.13075634837150574,
"learning_rate": 1.4512207637783138e-05,
"loss": 0.5407640337944031,
"step": 542
},
{
"epoch": 0.7438356164383562,
"grad_norm": 0.13585825264453888,
"learning_rate": 1.4492397141067888e-05,
"loss": 0.5683429837226868,
"step": 543
},
{
"epoch": 0.7452054794520548,
"grad_norm": 0.14631445705890656,
"learning_rate": 1.4472564531172558e-05,
"loss": 0.5532336235046387,
"step": 544
},
{
"epoch": 0.7465753424657534,
"grad_norm": 0.15100950002670288,
"learning_rate": 1.4452709905720314e-05,
"loss": 0.5505513548851013,
"step": 545
},
{
"epoch": 0.7479452054794521,
"grad_norm": 0.15109765529632568,
"learning_rate": 1.4432833362442708e-05,
"loss": 0.5818567276000977,
"step": 546
},
{
"epoch": 0.7493150684931507,
"grad_norm": 0.15869656205177307,
"learning_rate": 1.4412934999179169e-05,
"loss": 0.591871976852417,
"step": 547
},
{
"epoch": 0.7506849315068493,
"grad_norm": 0.13581599295139313,
"learning_rate": 1.4393014913876528e-05,
"loss": 0.5127702355384827,
"step": 548
},
{
"epoch": 0.7520547945205479,
"grad_norm": 0.1417117863893509,
"learning_rate": 1.4373073204588556e-05,
"loss": 0.5655680894851685,
"step": 549
},
{
"epoch": 0.7534246575342466,
"grad_norm": 0.15534058213233948,
"learning_rate": 1.4353109969475447e-05,
"loss": 0.577724814414978,
"step": 550
},
{
"epoch": 0.7547945205479452,
"grad_norm": 0.1513068974018097,
"learning_rate": 1.4333125306803362e-05,
"loss": 0.5496415495872498,
"step": 551
},
{
"epoch": 0.7561643835616438,
"grad_norm": 0.14230354130268097,
"learning_rate": 1.4313119314943933e-05,
"loss": 0.525494396686554,
"step": 552
},
{
"epoch": 0.7575342465753425,
"grad_norm": 0.13738034665584564,
"learning_rate": 1.4293092092373786e-05,
"loss": 0.519819974899292,
"step": 553
},
{
"epoch": 0.7589041095890411,
"grad_norm": 0.13675081729888916,
"learning_rate": 1.4273043737674048e-05,
"loss": 0.5584299564361572,
"step": 554
},
{
"epoch": 0.7602739726027398,
"grad_norm": 0.1386847048997879,
"learning_rate": 1.4252974349529871e-05,
"loss": 0.5083411335945129,
"step": 555
},
{
"epoch": 0.7616438356164383,
"grad_norm": 0.14538908004760742,
"learning_rate": 1.4232884026729927e-05,
"loss": 0.6105762720108032,
"step": 556
},
{
"epoch": 0.763013698630137,
"grad_norm": 0.15406113862991333,
"learning_rate": 1.4212772868165957e-05,
"loss": 0.5772764086723328,
"step": 557
},
{
"epoch": 0.7643835616438356,
"grad_norm": 0.16195538640022278,
"learning_rate": 1.419264097283225e-05,
"loss": 0.6081218719482422,
"step": 558
},
{
"epoch": 0.7657534246575343,
"grad_norm": 0.12385565787553787,
"learning_rate": 1.4172488439825167e-05,
"loss": 0.4937037229537964,
"step": 559
},
{
"epoch": 0.7671232876712328,
"grad_norm": 0.15183141827583313,
"learning_rate": 1.4152315368342661e-05,
"loss": 0.5616321563720703,
"step": 560
},
{
"epoch": 0.7684931506849315,
"grad_norm": 0.13611982762813568,
"learning_rate": 1.4132121857683782e-05,
"loss": 0.5686037540435791,
"step": 561
},
{
"epoch": 0.7698630136986301,
"grad_norm": 0.1663551926612854,
"learning_rate": 1.4111908007248188e-05,
"loss": 0.6417747735977173,
"step": 562
},
{
"epoch": 0.7712328767123288,
"grad_norm": 0.14242985844612122,
"learning_rate": 1.4091673916535655e-05,
"loss": 0.5676991939544678,
"step": 563
},
{
"epoch": 0.7726027397260274,
"grad_norm": 0.14829294383525848,
"learning_rate": 1.4071419685145587e-05,
"loss": 0.5915441513061523,
"step": 564
},
{
"epoch": 0.773972602739726,
"grad_norm": 0.15141542255878448,
"learning_rate": 1.4051145412776536e-05,
"loss": 0.5690219402313232,
"step": 565
},
{
"epoch": 0.7753424657534247,
"grad_norm": 0.13884322345256805,
"learning_rate": 1.4030851199225697e-05,
"loss": 0.5482669472694397,
"step": 566
},
{
"epoch": 0.7767123287671233,
"grad_norm": 0.141120046377182,
"learning_rate": 1.4010537144388416e-05,
"loss": 0.5446934103965759,
"step": 567
},
{
"epoch": 0.7780821917808219,
"grad_norm": 0.14071466028690338,
"learning_rate": 1.3990203348257722e-05,
"loss": 0.5528297424316406,
"step": 568
},
{
"epoch": 0.7794520547945205,
"grad_norm": 0.1462251991033554,
"learning_rate": 1.3969849910923803e-05,
"loss": 0.5360151529312134,
"step": 569
},
{
"epoch": 0.7808219178082192,
"grad_norm": 0.14809444546699524,
"learning_rate": 1.3949476932573531e-05,
"loss": 0.5697877407073975,
"step": 570
},
{
"epoch": 0.7821917808219178,
"grad_norm": 0.1515805721282959,
"learning_rate": 1.3929084513489972e-05,
"loss": 0.5751916170120239,
"step": 571
},
{
"epoch": 0.7835616438356164,
"grad_norm": 0.14923812448978424,
"learning_rate": 1.3908672754051877e-05,
"loss": 0.6014054417610168,
"step": 572
},
{
"epoch": 0.7849315068493151,
"grad_norm": 0.15840941667556763,
"learning_rate": 1.388824175473321e-05,
"loss": 0.5876986980438232,
"step": 573
},
{
"epoch": 0.7863013698630137,
"grad_norm": 0.14389535784721375,
"learning_rate": 1.386779161610262e-05,
"loss": 0.5680856704711914,
"step": 574
},
{
"epoch": 0.7876712328767124,
"grad_norm": 0.1493615210056305,
"learning_rate": 1.3847322438822988e-05,
"loss": 0.5676318407058716,
"step": 575
},
{
"epoch": 0.7890410958904109,
"grad_norm": 0.1369982659816742,
"learning_rate": 1.3826834323650899e-05,
"loss": 0.5286301970481873,
"step": 576
},
{
"epoch": 0.7904109589041096,
"grad_norm": 0.15035495162010193,
"learning_rate": 1.3806327371436159e-05,
"loss": 0.5285563468933105,
"step": 577
},
{
"epoch": 0.7917808219178082,
"grad_norm": 0.14841550588607788,
"learning_rate": 1.3785801683121296e-05,
"loss": 0.520169734954834,
"step": 578
},
{
"epoch": 0.7931506849315069,
"grad_norm": 0.14321400225162506,
"learning_rate": 1.3765257359741065e-05,
"loss": 0.5244216918945312,
"step": 579
},
{
"epoch": 0.7945205479452054,
"grad_norm": 0.15067681670188904,
"learning_rate": 1.3744694502421949e-05,
"loss": 0.6115273237228394,
"step": 580
},
{
"epoch": 0.7958904109589041,
"grad_norm": 0.14685063064098358,
"learning_rate": 1.372411321238166e-05,
"loss": 0.5616559982299805,
"step": 581
},
{
"epoch": 0.7972602739726027,
"grad_norm": 0.14456450939178467,
"learning_rate": 1.3703513590928647e-05,
"loss": 0.5625962018966675,
"step": 582
},
{
"epoch": 0.7986301369863014,
"grad_norm": 0.14497993886470795,
"learning_rate": 1.3682895739461588e-05,
"loss": 0.5844519138336182,
"step": 583
},
{
"epoch": 0.8,
"grad_norm": 0.1386929750442505,
"learning_rate": 1.3662259759468896e-05,
"loss": 0.5730550289154053,
"step": 584
},
{
"epoch": 0.8013698630136986,
"grad_norm": 0.13944153487682343,
"learning_rate": 1.3641605752528225e-05,
"loss": 0.5775356888771057,
"step": 585
},
{
"epoch": 0.8027397260273973,
"grad_norm": 0.14761903882026672,
"learning_rate": 1.3620933820305953e-05,
"loss": 0.5440607666969299,
"step": 586
},
{
"epoch": 0.8041095890410959,
"grad_norm": 0.14507032930850983,
"learning_rate": 1.3600244064556702e-05,
"loss": 0.5188732147216797,
"step": 587
},
{
"epoch": 0.8054794520547945,
"grad_norm": 0.14819754660129547,
"learning_rate": 1.3579536587122828e-05,
"loss": 0.5256346464157104,
"step": 588
},
{
"epoch": 0.8068493150684931,
"grad_norm": 0.14310747385025024,
"learning_rate": 1.3558811489933909e-05,
"loss": 0.5244719386100769,
"step": 589
},
{
"epoch": 0.8082191780821918,
"grad_norm": 0.1424933671951294,
"learning_rate": 1.3538068875006272e-05,
"loss": 0.520290732383728,
"step": 590
},
{
"epoch": 0.8095890410958904,
"grad_norm": 0.13971343636512756,
"learning_rate": 1.351730884444245e-05,
"loss": 0.5435692071914673,
"step": 591
},
{
"epoch": 0.810958904109589,
"grad_norm": 0.13619816303253174,
"learning_rate": 1.3496531500430723e-05,
"loss": 0.5121521949768066,
"step": 592
},
{
"epoch": 0.8123287671232877,
"grad_norm": 0.14922301471233368,
"learning_rate": 1.3475736945244575e-05,
"loss": 0.5281805396080017,
"step": 593
},
{
"epoch": 0.8136986301369863,
"grad_norm": 0.14446088671684265,
"learning_rate": 1.3454925281242225e-05,
"loss": 0.6063831448554993,
"step": 594
},
{
"epoch": 0.815068493150685,
"grad_norm": 0.14300888776779175,
"learning_rate": 1.3434096610866106e-05,
"loss": 0.5407198071479797,
"step": 595
},
{
"epoch": 0.8164383561643835,
"grad_norm": 0.13160032033920288,
"learning_rate": 1.3413251036642349e-05,
"loss": 0.5404701232910156,
"step": 596
},
{
"epoch": 0.8178082191780822,
"grad_norm": 0.13810713589191437,
"learning_rate": 1.3392388661180303e-05,
"loss": 0.5198618769645691,
"step": 597
},
{
"epoch": 0.8191780821917808,
"grad_norm": 0.14353394508361816,
"learning_rate": 1.3371509587172023e-05,
"loss": 0.513703465461731,
"step": 598
},
{
"epoch": 0.8205479452054795,
"grad_norm": 0.1333569884300232,
"learning_rate": 1.3350613917391749e-05,
"loss": 0.5600924491882324,
"step": 599
},
{
"epoch": 0.821917808219178,
"grad_norm": 0.15609651803970337,
"learning_rate": 1.3329701754695412e-05,
"loss": 0.6079143285751343,
"step": 600
},
{
"epoch": 0.8232876712328767,
"grad_norm": 0.13941015303134918,
"learning_rate": 1.330877320202014e-05,
"loss": 0.5294625759124756,
"step": 601
},
{
"epoch": 0.8246575342465754,
"grad_norm": 0.14244301617145538,
"learning_rate": 1.3287828362383724e-05,
"loss": 0.6567379236221313,
"step": 602
},
{
"epoch": 0.826027397260274,
"grad_norm": 0.1582559496164322,
"learning_rate": 1.3266867338884131e-05,
"loss": 0.5663045644760132,
"step": 603
},
{
"epoch": 0.8273972602739726,
"grad_norm": 0.16126839816570282,
"learning_rate": 1.3245890234698993e-05,
"loss": 0.5470207333564758,
"step": 604
},
{
"epoch": 0.8287671232876712,
"grad_norm": 0.14516150951385498,
"learning_rate": 1.322489715308509e-05,
"loss": 0.5473780035972595,
"step": 605
},
{
"epoch": 0.8301369863013699,
"grad_norm": 0.13455289602279663,
"learning_rate": 1.3203888197377857e-05,
"loss": 0.5864403247833252,
"step": 606
},
{
"epoch": 0.8315068493150685,
"grad_norm": 0.1354503482580185,
"learning_rate": 1.318286347099086e-05,
"loss": 0.5317245721817017,
"step": 607
},
{
"epoch": 0.8328767123287671,
"grad_norm": 0.14837303757667542,
"learning_rate": 1.3161823077415299e-05,
"loss": 0.5762374401092529,
"step": 608
},
{
"epoch": 0.8342465753424657,
"grad_norm": 0.14379873871803284,
"learning_rate": 1.314076712021949e-05,
"loss": 0.5885000228881836,
"step": 609
},
{
"epoch": 0.8356164383561644,
"grad_norm": 0.1361761838197708,
"learning_rate": 1.3119695703048358e-05,
"loss": 0.43967384099960327,
"step": 610
},
{
"epoch": 0.836986301369863,
"grad_norm": 0.13545174896717072,
"learning_rate": 1.3098608929622934e-05,
"loss": 0.4924185276031494,
"step": 611
},
{
"epoch": 0.8383561643835616,
"grad_norm": 0.13373355567455292,
"learning_rate": 1.3077506903739829e-05,
"loss": 0.536264181137085,
"step": 612
},
{
"epoch": 0.8397260273972603,
"grad_norm": 0.15098173916339874,
"learning_rate": 1.305638972927074e-05,
"loss": 0.6109551787376404,
"step": 613
},
{
"epoch": 0.8410958904109589,
"grad_norm": 0.13542239367961884,
"learning_rate": 1.3035257510161925e-05,
"loss": 0.4766671657562256,
"step": 614
},
{
"epoch": 0.8424657534246576,
"grad_norm": 0.1498875766992569,
"learning_rate": 1.30141103504337e-05,
"loss": 0.598709225654602,
"step": 615
},
{
"epoch": 0.8438356164383561,
"grad_norm": 0.14445774257183075,
"learning_rate": 1.2992948354179919e-05,
"loss": 0.5338215827941895,
"step": 616
},
{
"epoch": 0.8452054794520548,
"grad_norm": 0.1474190205335617,
"learning_rate": 1.297177162556748e-05,
"loss": 0.518726110458374,
"step": 617
},
{
"epoch": 0.8465753424657534,
"grad_norm": 0.14172446727752686,
"learning_rate": 1.2950580268835784e-05,
"loss": 0.5487723350524902,
"step": 618
},
{
"epoch": 0.8479452054794521,
"grad_norm": 0.14646317064762115,
"learning_rate": 1.2929374388296241e-05,
"loss": 0.545096218585968,
"step": 619
},
{
"epoch": 0.8493150684931506,
"grad_norm": 0.14491736888885498,
"learning_rate": 1.2908154088331757e-05,
"loss": 0.5572469830513,
"step": 620
},
{
"epoch": 0.8506849315068493,
"grad_norm": 0.14330042898654938,
"learning_rate": 1.2886919473396212e-05,
"loss": 0.5331091284751892,
"step": 621
},
{
"epoch": 0.852054794520548,
"grad_norm": 0.13608761131763458,
"learning_rate": 1.2865670648013951e-05,
"loss": 0.5101725459098816,
"step": 622
},
{
"epoch": 0.8534246575342466,
"grad_norm": 0.15225724875926971,
"learning_rate": 1.2844407716779266e-05,
"loss": 0.5518755912780762,
"step": 623
},
{
"epoch": 0.8547945205479452,
"grad_norm": 0.14294537901878357,
"learning_rate": 1.2823130784355882e-05,
"loss": 0.48980948328971863,
"step": 624
},
{
"epoch": 0.8561643835616438,
"grad_norm": 0.13224340975284576,
"learning_rate": 1.2801839955476444e-05,
"loss": 0.4973132610321045,
"step": 625
},
{
"epoch": 0.8575342465753425,
"grad_norm": 0.1523551344871521,
"learning_rate": 1.2780535334942011e-05,
"loss": 0.5488629341125488,
"step": 626
},
{
"epoch": 0.8589041095890411,
"grad_norm": 0.12789522111415863,
"learning_rate": 1.2759217027621507e-05,
"loss": 0.5876277685165405,
"step": 627
},
{
"epoch": 0.8602739726027397,
"grad_norm": 0.12671145796775818,
"learning_rate": 1.2737885138451246e-05,
"loss": 0.5252135992050171,
"step": 628
},
{
"epoch": 0.8616438356164383,
"grad_norm": 0.13611386716365814,
"learning_rate": 1.2716539772434389e-05,
"loss": 0.5474650859832764,
"step": 629
},
{
"epoch": 0.863013698630137,
"grad_norm": 0.13836732506752014,
"learning_rate": 1.2695181034640435e-05,
"loss": 0.5324945449829102,
"step": 630
},
{
"epoch": 0.8643835616438356,
"grad_norm": 0.13409800827503204,
"learning_rate": 1.267380903020471e-05,
"loss": 0.5111215710639954,
"step": 631
},
{
"epoch": 0.8657534246575342,
"grad_norm": 0.14573852717876434,
"learning_rate": 1.2652423864327833e-05,
"loss": 0.551156759262085,
"step": 632
},
{
"epoch": 0.8671232876712329,
"grad_norm": 0.1373865008354187,
"learning_rate": 1.2631025642275212e-05,
"loss": 0.5151498317718506,
"step": 633
},
{
"epoch": 0.8684931506849315,
"grad_norm": 0.15578457713127136,
"learning_rate": 1.2609614469376532e-05,
"loss": 0.5745099782943726,
"step": 634
},
{
"epoch": 0.8698630136986302,
"grad_norm": 0.14487750828266144,
"learning_rate": 1.2588190451025209e-05,
"loss": 0.591609001159668,
"step": 635
},
{
"epoch": 0.8712328767123287,
"grad_norm": 0.1382559835910797,
"learning_rate": 1.2566753692677902e-05,
"loss": 0.5301614999771118,
"step": 636
},
{
"epoch": 0.8726027397260274,
"grad_norm": 0.1299545019865036,
"learning_rate": 1.2545304299853977e-05,
"loss": 0.4942835569381714,
"step": 637
},
{
"epoch": 0.873972602739726,
"grad_norm": 0.13769906759262085,
"learning_rate": 1.2523842378134991e-05,
"loss": 0.5325908660888672,
"step": 638
},
{
"epoch": 0.8753424657534247,
"grad_norm": 0.13713912665843964,
"learning_rate": 1.2502368033164176e-05,
"loss": 0.5291322469711304,
"step": 639
},
{
"epoch": 0.8767123287671232,
"grad_norm": 0.14469754695892334,
"learning_rate": 1.2480881370645906e-05,
"loss": 0.5346277952194214,
"step": 640
},
{
"epoch": 0.8780821917808219,
"grad_norm": 0.1554187834262848,
"learning_rate": 1.2459382496345199e-05,
"loss": 0.5714985728263855,
"step": 641
},
{
"epoch": 0.8794520547945206,
"grad_norm": 0.14312691986560822,
"learning_rate": 1.2437871516087174e-05,
"loss": 0.543634831905365,
"step": 642
},
{
"epoch": 0.8808219178082192,
"grad_norm": 0.14826460182666779,
"learning_rate": 1.2416348535756543e-05,
"loss": 0.5458000898361206,
"step": 643
},
{
"epoch": 0.8821917808219178,
"grad_norm": 0.144753098487854,
"learning_rate": 1.2394813661297093e-05,
"loss": 0.5427224636077881,
"step": 644
},
{
"epoch": 0.8835616438356164,
"grad_norm": 0.17040959000587463,
"learning_rate": 1.2373266998711152e-05,
"loss": 0.6165363192558289,
"step": 645
},
{
"epoch": 0.8849315068493151,
"grad_norm": 0.1435997039079666,
"learning_rate": 1.2351708654059065e-05,
"loss": 0.49909958243370056,
"step": 646
},
{
"epoch": 0.8863013698630137,
"grad_norm": 0.139425128698349,
"learning_rate": 1.2330138733458693e-05,
"loss": 0.5313612222671509,
"step": 647
},
{
"epoch": 0.8876712328767123,
"grad_norm": 0.13428698480129242,
"learning_rate": 1.2308557343084881e-05,
"loss": 0.5440371632575989,
"step": 648
},
{
"epoch": 0.8890410958904109,
"grad_norm": 0.14464199542999268,
"learning_rate": 1.2286964589168917e-05,
"loss": 0.6058000326156616,
"step": 649
},
{
"epoch": 0.8904109589041096,
"grad_norm": 0.13464415073394775,
"learning_rate": 1.2265360577998038e-05,
"loss": 0.5244266986846924,
"step": 650
},
{
"epoch": 0.8917808219178082,
"grad_norm": 0.17547836899757385,
"learning_rate": 1.2243745415914882e-05,
"loss": 0.5788463354110718,
"step": 651
},
{
"epoch": 0.8931506849315068,
"grad_norm": 0.1391308307647705,
"learning_rate": 1.2222119209316984e-05,
"loss": 0.5580483078956604,
"step": 652
},
{
"epoch": 0.8945205479452055,
"grad_norm": 0.15701870620250702,
"learning_rate": 1.220048206465625e-05,
"loss": 0.6042839288711548,
"step": 653
},
{
"epoch": 0.8958904109589041,
"grad_norm": 0.12679646909236908,
"learning_rate": 1.2178834088438404e-05,
"loss": 0.5429613590240479,
"step": 654
},
{
"epoch": 0.8972602739726028,
"grad_norm": 0.13499225676059723,
"learning_rate": 1.2157175387222516e-05,
"loss": 0.576964795589447,
"step": 655
},
{
"epoch": 0.8986301369863013,
"grad_norm": 0.12962238490581512,
"learning_rate": 1.2135506067620421e-05,
"loss": 0.4955563247203827,
"step": 656
},
{
"epoch": 0.9,
"grad_norm": 0.14396265149116516,
"learning_rate": 1.2113826236296245e-05,
"loss": 0.5154756903648376,
"step": 657
},
{
"epoch": 0.9013698630136986,
"grad_norm": 0.1490243673324585,
"learning_rate": 1.209213599996584e-05,
"loss": 0.6035890579223633,
"step": 658
},
{
"epoch": 0.9027397260273973,
"grad_norm": 0.15239585936069489,
"learning_rate": 1.2070435465396283e-05,
"loss": 0.587848424911499,
"step": 659
},
{
"epoch": 0.9041095890410958,
"grad_norm": 0.1464146375656128,
"learning_rate": 1.2048724739405337e-05,
"loss": 0.6095236539840698,
"step": 660
},
{
"epoch": 0.9054794520547945,
"grad_norm": 0.12767134606838226,
"learning_rate": 1.2027003928860936e-05,
"loss": 0.4720427691936493,
"step": 661
},
{
"epoch": 0.9068493150684932,
"grad_norm": 0.14553502202033997,
"learning_rate": 1.2005273140680656e-05,
"loss": 0.5657269358634949,
"step": 662
},
{
"epoch": 0.9082191780821918,
"grad_norm": 0.13808585703372955,
"learning_rate": 1.1983532481831179e-05,
"loss": 0.4915253818035126,
"step": 663
},
{
"epoch": 0.9095890410958904,
"grad_norm": 0.14350081980228424,
"learning_rate": 1.196178205932778e-05,
"loss": 0.5666084289550781,
"step": 664
},
{
"epoch": 0.910958904109589,
"grad_norm": 0.142891526222229,
"learning_rate": 1.1940021980233784e-05,
"loss": 0.5144902467727661,
"step": 665
},
{
"epoch": 0.9123287671232877,
"grad_norm": 0.13920342922210693,
"learning_rate": 1.1918252351660066e-05,
"loss": 0.5250864028930664,
"step": 666
},
{
"epoch": 0.9136986301369863,
"grad_norm": 0.13702407479286194,
"learning_rate": 1.1896473280764498e-05,
"loss": 0.49932771921157837,
"step": 667
},
{
"epoch": 0.915068493150685,
"grad_norm": 0.15072178840637207,
"learning_rate": 1.187468487475142e-05,
"loss": 0.5834764242172241,
"step": 668
},
{
"epoch": 0.9164383561643835,
"grad_norm": 0.1453569233417511,
"learning_rate": 1.1852887240871145e-05,
"loss": 0.5249761343002319,
"step": 669
},
{
"epoch": 0.9178082191780822,
"grad_norm": 0.1490030437707901,
"learning_rate": 1.1831080486419388e-05,
"loss": 0.5667267441749573,
"step": 670
},
{
"epoch": 0.9191780821917809,
"grad_norm": 0.1355256885290146,
"learning_rate": 1.1809264718736775e-05,
"loss": 0.5361207723617554,
"step": 671
},
{
"epoch": 0.9205479452054794,
"grad_norm": 0.1491296887397766,
"learning_rate": 1.1787440045208287e-05,
"loss": 0.5815398693084717,
"step": 672
},
{
"epoch": 0.9219178082191781,
"grad_norm": 0.1443261355161667,
"learning_rate": 1.1765606573262745e-05,
"loss": 0.5183050036430359,
"step": 673
},
{
"epoch": 0.9232876712328767,
"grad_norm": 0.13231635093688965,
"learning_rate": 1.1743764410372283e-05,
"loss": 0.570467472076416,
"step": 674
},
{
"epoch": 0.9246575342465754,
"grad_norm": 0.1384153813123703,
"learning_rate": 1.1721913664051814e-05,
"loss": 0.544106125831604,
"step": 675
},
{
"epoch": 0.9260273972602739,
"grad_norm": 0.15208759903907776,
"learning_rate": 1.1700054441858497e-05,
"loss": 0.5341042876243591,
"step": 676
},
{
"epoch": 0.9273972602739726,
"grad_norm": 0.14940430223941803,
"learning_rate": 1.1678186851391218e-05,
"loss": 0.5699517726898193,
"step": 677
},
{
"epoch": 0.9287671232876712,
"grad_norm": 0.14947940409183502,
"learning_rate": 1.165631100029005e-05,
"loss": 0.5162645578384399,
"step": 678
},
{
"epoch": 0.9301369863013699,
"grad_norm": 0.12951557338237762,
"learning_rate": 1.1634426996235732e-05,
"loss": 0.5086703300476074,
"step": 679
},
{
"epoch": 0.9315068493150684,
"grad_norm": 0.14504235982894897,
"learning_rate": 1.1612534946949134e-05,
"loss": 0.475946307182312,
"step": 680
},
{
"epoch": 0.9328767123287671,
"grad_norm": 0.13910368084907532,
"learning_rate": 1.1590634960190722e-05,
"loss": 0.5487220883369446,
"step": 681
},
{
"epoch": 0.9342465753424658,
"grad_norm": 0.15137788653373718,
"learning_rate": 1.1568727143760042e-05,
"loss": 0.5802314281463623,
"step": 682
},
{
"epoch": 0.9356164383561644,
"grad_norm": 0.13175715506076813,
"learning_rate": 1.154681160549517e-05,
"loss": 0.5751426219940186,
"step": 683
},
{
"epoch": 0.936986301369863,
"grad_norm": 0.1453649252653122,
"learning_rate": 1.15248884532722e-05,
"loss": 0.563701868057251,
"step": 684
},
{
"epoch": 0.9383561643835616,
"grad_norm": 0.14549468457698822,
"learning_rate": 1.1502957795004706e-05,
"loss": 0.5319613218307495,
"step": 685
},
{
"epoch": 0.9397260273972603,
"grad_norm": 0.14660926163196564,
"learning_rate": 1.1481019738643202e-05,
"loss": 0.5583406686782837,
"step": 686
},
{
"epoch": 0.9410958904109589,
"grad_norm": 0.15349538624286652,
"learning_rate": 1.1459074392174619e-05,
"loss": 0.5592145919799805,
"step": 687
},
{
"epoch": 0.9424657534246575,
"grad_norm": 0.15075471997261047,
"learning_rate": 1.1437121863621776e-05,
"loss": 0.5834563374519348,
"step": 688
},
{
"epoch": 0.9438356164383561,
"grad_norm": 0.13851697742938995,
"learning_rate": 1.1415162261042849e-05,
"loss": 0.5394259691238403,
"step": 689
},
{
"epoch": 0.9452054794520548,
"grad_norm": 0.13821455836296082,
"learning_rate": 1.1393195692530823e-05,
"loss": 0.5597106218338013,
"step": 690
},
{
"epoch": 0.9465753424657535,
"grad_norm": 0.13941477239131927,
"learning_rate": 1.1371222266212983e-05,
"loss": 0.5378660559654236,
"step": 691
},
{
"epoch": 0.947945205479452,
"grad_norm": 0.15957188606262207,
"learning_rate": 1.134924209025036e-05,
"loss": 0.559799075126648,
"step": 692
},
{
"epoch": 0.9493150684931507,
"grad_norm": 0.14969097077846527,
"learning_rate": 1.1327255272837221e-05,
"loss": 0.5599546432495117,
"step": 693
},
{
"epoch": 0.9506849315068493,
"grad_norm": 0.1421581506729126,
"learning_rate": 1.130526192220052e-05,
"loss": 0.5460138320922852,
"step": 694
},
{
"epoch": 0.952054794520548,
"grad_norm": 0.1384434700012207,
"learning_rate": 1.128326214659936e-05,
"loss": 0.5014693140983582,
"step": 695
},
{
"epoch": 0.9534246575342465,
"grad_norm": 0.13422176241874695,
"learning_rate": 1.1261256054324488e-05,
"loss": 0.4789910912513733,
"step": 696
},
{
"epoch": 0.9547945205479452,
"grad_norm": 0.15614454448223114,
"learning_rate": 1.1239243753697728e-05,
"loss": 0.5928708910942078,
"step": 697
},
{
"epoch": 0.9561643835616438,
"grad_norm": 0.13170309364795685,
"learning_rate": 1.1217225353071473e-05,
"loss": 0.5444527268409729,
"step": 698
},
{
"epoch": 0.9575342465753425,
"grad_norm": 0.143415629863739,
"learning_rate": 1.1195200960828138e-05,
"loss": 0.5303205847740173,
"step": 699
},
{
"epoch": 0.958904109589041,
"grad_norm": 0.14177711308002472,
"learning_rate": 1.1173170685379635e-05,
"loss": 0.5031203627586365,
"step": 700
},
{
"epoch": 0.9602739726027397,
"grad_norm": 0.15743711590766907,
"learning_rate": 1.115113463516683e-05,
"loss": 0.5337619781494141,
"step": 701
},
{
"epoch": 0.9616438356164384,
"grad_norm": 0.14667122066020966,
"learning_rate": 1.1129092918659019e-05,
"loss": 0.48960891366004944,
"step": 702
},
{
"epoch": 0.963013698630137,
"grad_norm": 0.1391826868057251,
"learning_rate": 1.1107045644353388e-05,
"loss": 0.5426533222198486,
"step": 703
},
{
"epoch": 0.9643835616438357,
"grad_norm": 0.15035362541675568,
"learning_rate": 1.1084992920774478e-05,
"loss": 0.4937519431114197,
"step": 704
},
{
"epoch": 0.9657534246575342,
"grad_norm": 0.14533023536205292,
"learning_rate": 1.1062934856473655e-05,
"loss": 0.5189172029495239,
"step": 705
},
{
"epoch": 0.9671232876712329,
"grad_norm": 0.12609876692295074,
"learning_rate": 1.1040871560028576e-05,
"loss": 0.5009540915489197,
"step": 706
},
{
"epoch": 0.9684931506849315,
"grad_norm": 0.1392073929309845,
"learning_rate": 1.1018803140042651e-05,
"loss": 0.5657528638839722,
"step": 707
},
{
"epoch": 0.9698630136986301,
"grad_norm": 0.1347883641719818,
"learning_rate": 1.099672970514451e-05,
"loss": 0.5511807799339294,
"step": 708
},
{
"epoch": 0.9712328767123287,
"grad_norm": 0.15112189948558807,
"learning_rate": 1.0974651363987465e-05,
"loss": 0.5117864608764648,
"step": 709
},
{
"epoch": 0.9726027397260274,
"grad_norm": 0.1492203027009964,
"learning_rate": 1.0952568225248983e-05,
"loss": 0.5602498054504395,
"step": 710
},
{
"epoch": 0.9739726027397261,
"grad_norm": 0.13084228336811066,
"learning_rate": 1.0930480397630146e-05,
"loss": 0.503504753112793,
"step": 711
},
{
"epoch": 0.9753424657534246,
"grad_norm": 0.12569855153560638,
"learning_rate": 1.0908387989855114e-05,
"loss": 0.49362748861312866,
"step": 712
},
{
"epoch": 0.9767123287671233,
"grad_norm": 0.14552345871925354,
"learning_rate": 1.0886291110670592e-05,
"loss": 0.5039536952972412,
"step": 713
},
{
"epoch": 0.9780821917808219,
"grad_norm": 0.13342419266700745,
"learning_rate": 1.0864189868845296e-05,
"loss": 0.5050915479660034,
"step": 714
},
{
"epoch": 0.9794520547945206,
"grad_norm": 0.15216471254825592,
"learning_rate": 1.0842084373169417e-05,
"loss": 0.5598911046981812,
"step": 715
},
{
"epoch": 0.9808219178082191,
"grad_norm": 0.12561655044555664,
"learning_rate": 1.0819974732454086e-05,
"loss": 0.47096988558769226,
"step": 716
},
{
"epoch": 0.9821917808219178,
"grad_norm": 0.1414027214050293,
"learning_rate": 1.0797861055530832e-05,
"loss": 0.5470199584960938,
"step": 717
},
{
"epoch": 0.9835616438356164,
"grad_norm": 0.14046591520309448,
"learning_rate": 1.077574345125106e-05,
"loss": 0.5058327913284302,
"step": 718
},
{
"epoch": 0.9849315068493151,
"grad_norm": 0.13157160580158234,
"learning_rate": 1.0753622028485498e-05,
"loss": 0.4962088465690613,
"step": 719
},
{
"epoch": 0.9863013698630136,
"grad_norm": 0.1477247178554535,
"learning_rate": 1.0731496896123676e-05,
"loss": 0.611939549446106,
"step": 720
},
{
"epoch": 0.9876712328767123,
"grad_norm": 0.1309177726507187,
"learning_rate": 1.0709368163073385e-05,
"loss": 0.5082958936691284,
"step": 721
},
{
"epoch": 0.989041095890411,
"grad_norm": 0.13591881096363068,
"learning_rate": 1.0687235938260134e-05,
"loss": 0.5050320625305176,
"step": 722
},
{
"epoch": 0.9904109589041096,
"grad_norm": 0.1385071724653244,
"learning_rate": 1.0665100330626625e-05,
"loss": 0.5245853066444397,
"step": 723
},
{
"epoch": 0.9917808219178083,
"grad_norm": 0.14319604635238647,
"learning_rate": 1.0642961449132203e-05,
"loss": 0.47970306873321533,
"step": 724
},
{
"epoch": 0.9931506849315068,
"grad_norm": 0.131264328956604,
"learning_rate": 1.062081940275234e-05,
"loss": 0.4639028310775757,
"step": 725
},
{
"epoch": 0.9945205479452055,
"grad_norm": 0.14266684651374817,
"learning_rate": 1.0598674300478085e-05,
"loss": 0.521947979927063,
"step": 726
},
{
"epoch": 0.9958904109589041,
"grad_norm": 0.13719575107097626,
"learning_rate": 1.0576526251315515e-05,
"loss": 0.5097373723983765,
"step": 727
},
{
"epoch": 0.9972602739726028,
"grad_norm": 0.13589054346084595,
"learning_rate": 1.055437536428523e-05,
"loss": 0.5200362205505371,
"step": 728
},
{
"epoch": 0.9986301369863013,
"grad_norm": 0.14577971398830414,
"learning_rate": 1.0532221748421786e-05,
"loss": 0.5406281352043152,
"step": 729
},
{
"epoch": 1.0,
"grad_norm": 0.1317002773284912,
"learning_rate": 1.0510065512773184e-05,
"loss": 0.4858514964580536,
"step": 730
},
{
"epoch": 1.0013698630136987,
"grad_norm": 0.13495707511901855,
"learning_rate": 1.0487906766400313e-05,
"loss": 0.3737938404083252,
"step": 731
},
{
"epoch": 1.0027397260273974,
"grad_norm": 0.1421145498752594,
"learning_rate": 1.0465745618376417e-05,
"loss": 0.4118575155735016,
"step": 732
},
{
"epoch": 1.0041095890410958,
"grad_norm": 0.1352914422750473,
"learning_rate": 1.0443582177786564e-05,
"loss": 0.39527586102485657,
"step": 733
},
{
"epoch": 1.0054794520547945,
"grad_norm": 0.12854884564876556,
"learning_rate": 1.042141655372712e-05,
"loss": 0.3931320905685425,
"step": 734
},
{
"epoch": 1.0068493150684932,
"grad_norm": 0.14797155559062958,
"learning_rate": 1.0399248855305178e-05,
"loss": 0.3988172113895416,
"step": 735
},
{
"epoch": 1.0082191780821919,
"grad_norm": 0.13468636572360992,
"learning_rate": 1.0377079191638054e-05,
"loss": 0.39413654804229736,
"step": 736
},
{
"epoch": 1.0095890410958903,
"grad_norm": 0.1381807178258896,
"learning_rate": 1.0354907671852733e-05,
"loss": 0.37008973956108093,
"step": 737
},
{
"epoch": 1.010958904109589,
"grad_norm": 0.13726525008678436,
"learning_rate": 1.0332734405085343e-05,
"loss": 0.35282236337661743,
"step": 738
},
{
"epoch": 1.0123287671232877,
"grad_norm": 0.14333774149417877,
"learning_rate": 1.0310559500480603e-05,
"loss": 0.3581743836402893,
"step": 739
},
{
"epoch": 1.0136986301369864,
"grad_norm": 0.1522689312696457,
"learning_rate": 1.0288383067191305e-05,
"loss": 0.3681914210319519,
"step": 740
},
{
"epoch": 1.015068493150685,
"grad_norm": 0.17875906825065613,
"learning_rate": 1.026620521437775e-05,
"loss": 0.40607860684394836,
"step": 741
},
{
"epoch": 1.0164383561643835,
"grad_norm": 0.1594308316707611,
"learning_rate": 1.0244026051207236e-05,
"loss": 0.32388749718666077,
"step": 742
},
{
"epoch": 1.0178082191780822,
"grad_norm": 0.14830395579338074,
"learning_rate": 1.0221845686853517e-05,
"loss": 0.35188883543014526,
"step": 743
},
{
"epoch": 1.0191780821917809,
"grad_norm": 0.1651025116443634,
"learning_rate": 1.0199664230496247e-05,
"loss": 0.3737831711769104,
"step": 744
},
{
"epoch": 1.0205479452054795,
"grad_norm": 0.15182435512542725,
"learning_rate": 1.0177481791320463e-05,
"loss": 0.34954410791397095,
"step": 745
},
{
"epoch": 1.021917808219178,
"grad_norm": 0.16264082491397858,
"learning_rate": 1.0155298478516036e-05,
"loss": 0.35713765025138855,
"step": 746
},
{
"epoch": 1.0232876712328767,
"grad_norm": 0.1404428780078888,
"learning_rate": 1.013311440127714e-05,
"loss": 0.33602964878082275,
"step": 747
},
{
"epoch": 1.0246575342465754,
"grad_norm": 0.14297620952129364,
"learning_rate": 1.0110929668801712e-05,
"loss": 0.359991192817688,
"step": 748
},
{
"epoch": 1.026027397260274,
"grad_norm": 0.1350749284029007,
"learning_rate": 1.008874439029091e-05,
"loss": 0.3239899277687073,
"step": 749
},
{
"epoch": 1.0273972602739727,
"grad_norm": 0.16016657650470734,
"learning_rate": 1.006655867494859e-05,
"loss": 0.3807777464389801,
"step": 750
},
{
"epoch": 1.0287671232876712,
"grad_norm": 0.15092220902442932,
"learning_rate": 1.0044372631980742e-05,
"loss": 0.35382795333862305,
"step": 751
},
{
"epoch": 1.0301369863013699,
"grad_norm": 0.15390557050704956,
"learning_rate": 1.0022186370594982e-05,
"loss": 0.36204713582992554,
"step": 752
},
{
"epoch": 1.0315068493150685,
"grad_norm": 0.13931438326835632,
"learning_rate": 1e-05,
"loss": 0.3561636209487915,
"step": 753
},
{
"epoch": 1.0328767123287672,
"grad_norm": 0.1428987830877304,
"learning_rate": 9.977813629405018e-06,
"loss": 0.3381158113479614,
"step": 754
},
{
"epoch": 1.0342465753424657,
"grad_norm": 0.15198323130607605,
"learning_rate": 9.95562736801926e-06,
"loss": 0.38106226921081543,
"step": 755
},
{
"epoch": 1.0356164383561643,
"grad_norm": 0.14457497000694275,
"learning_rate": 9.933441325051414e-06,
"loss": 0.3454380929470062,
"step": 756
},
{
"epoch": 1.036986301369863,
"grad_norm": 0.13807366788387299,
"learning_rate": 9.911255609709089e-06,
"loss": 0.32895246148109436,
"step": 757
},
{
"epoch": 1.0383561643835617,
"grad_norm": 0.13656342029571533,
"learning_rate": 9.889070331198293e-06,
"loss": 0.29782530665397644,
"step": 758
},
{
"epoch": 1.0397260273972604,
"grad_norm": 0.14839798212051392,
"learning_rate": 9.866885598722865e-06,
"loss": 0.3281187117099762,
"step": 759
},
{
"epoch": 1.0410958904109588,
"grad_norm": 0.13550812005996704,
"learning_rate": 9.844701521483967e-06,
"loss": 0.33477783203125,
"step": 760
},
{
"epoch": 1.0424657534246575,
"grad_norm": 0.15942195057868958,
"learning_rate": 9.822518208679542e-06,
"loss": 0.3608595132827759,
"step": 761
},
{
"epoch": 1.0438356164383562,
"grad_norm": 0.1583387702703476,
"learning_rate": 9.800335769503756e-06,
"loss": 0.3741905987262726,
"step": 762
},
{
"epoch": 1.0452054794520549,
"grad_norm": 0.14900487661361694,
"learning_rate": 9.778154313146485e-06,
"loss": 0.3473421335220337,
"step": 763
},
{
"epoch": 1.0465753424657533,
"grad_norm": 0.1592230647802353,
"learning_rate": 9.755973948792764e-06,
"loss": 0.3724527955055237,
"step": 764
},
{
"epoch": 1.047945205479452,
"grad_norm": 0.14690528810024261,
"learning_rate": 9.733794785622254e-06,
"loss": 0.3439611494541168,
"step": 765
},
{
"epoch": 1.0493150684931507,
"grad_norm": 0.1537773311138153,
"learning_rate": 9.711616932808698e-06,
"loss": 0.3581935167312622,
"step": 766
},
{
"epoch": 1.0506849315068494,
"grad_norm": 0.1587516963481903,
"learning_rate": 9.689440499519395e-06,
"loss": 0.3731722831726074,
"step": 767
},
{
"epoch": 1.0520547945205478,
"grad_norm": 0.15445035696029663,
"learning_rate": 9.667265594914662e-06,
"loss": 0.3295559585094452,
"step": 768
},
{
"epoch": 1.0534246575342465,
"grad_norm": 0.15352846682071686,
"learning_rate": 9.64509232814727e-06,
"loss": 0.36118239164352417,
"step": 769
},
{
"epoch": 1.0547945205479452,
"grad_norm": 0.16287478804588318,
"learning_rate": 9.622920808361948e-06,
"loss": 0.3689691126346588,
"step": 770
},
{
"epoch": 1.0561643835616439,
"grad_norm": 0.16106897592544556,
"learning_rate": 9.600751144694827e-06,
"loss": 0.363809198141098,
"step": 771
},
{
"epoch": 1.0575342465753426,
"grad_norm": 0.159645676612854,
"learning_rate": 9.578583446272885e-06,
"loss": 0.35414397716522217,
"step": 772
},
{
"epoch": 1.058904109589041,
"grad_norm": 0.13788153231143951,
"learning_rate": 9.556417822213436e-06,
"loss": 0.3615794777870178,
"step": 773
},
{
"epoch": 1.0602739726027397,
"grad_norm": 0.1447700411081314,
"learning_rate": 9.534254381623588e-06,
"loss": 0.3441850244998932,
"step": 774
},
{
"epoch": 1.0616438356164384,
"grad_norm": 0.14863401651382446,
"learning_rate": 9.51209323359969e-06,
"loss": 0.3699710965156555,
"step": 775
},
{
"epoch": 1.063013698630137,
"grad_norm": 0.16527967154979706,
"learning_rate": 9.489934487226816e-06,
"loss": 0.39048323035240173,
"step": 776
},
{
"epoch": 1.0643835616438355,
"grad_norm": 0.14739027619361877,
"learning_rate": 9.467778251578217e-06,
"loss": 0.32588720321655273,
"step": 777
},
{
"epoch": 1.0657534246575342,
"grad_norm": 0.14020374417304993,
"learning_rate": 9.445624635714773e-06,
"loss": 0.34290170669555664,
"step": 778
},
{
"epoch": 1.0671232876712329,
"grad_norm": 0.15601703524589539,
"learning_rate": 9.423473748684487e-06,
"loss": 0.3982515037059784,
"step": 779
},
{
"epoch": 1.0684931506849316,
"grad_norm": 0.15038882195949554,
"learning_rate": 9.401325699521922e-06,
"loss": 0.36383989453315735,
"step": 780
},
{
"epoch": 1.0698630136986302,
"grad_norm": 0.13526614010334015,
"learning_rate": 9.379180597247661e-06,
"loss": 0.35562628507614136,
"step": 781
},
{
"epoch": 1.0712328767123287,
"grad_norm": 0.15694350004196167,
"learning_rate": 9.357038550867799e-06,
"loss": 0.3323221802711487,
"step": 782
},
{
"epoch": 1.0726027397260274,
"grad_norm": 0.14334630966186523,
"learning_rate": 9.334899669373379e-06,
"loss": 0.3451099395751953,
"step": 783
},
{
"epoch": 1.073972602739726,
"grad_norm": 0.1410481035709381,
"learning_rate": 9.312764061739869e-06,
"loss": 0.3644659221172333,
"step": 784
},
{
"epoch": 1.0753424657534247,
"grad_norm": 0.13995599746704102,
"learning_rate": 9.290631836926616e-06,
"loss": 0.33525389432907104,
"step": 785
},
{
"epoch": 1.0767123287671232,
"grad_norm": 0.13992056250572205,
"learning_rate": 9.268503103876324e-06,
"loss": 0.3333061933517456,
"step": 786
},
{
"epoch": 1.0780821917808219,
"grad_norm": 0.15618863701820374,
"learning_rate": 9.246377971514504e-06,
"loss": 0.3487616777420044,
"step": 787
},
{
"epoch": 1.0794520547945206,
"grad_norm": 0.13778771460056305,
"learning_rate": 9.224256548748942e-06,
"loss": 0.3259185254573822,
"step": 788
},
{
"epoch": 1.0808219178082192,
"grad_norm": 0.16598646342754364,
"learning_rate": 9.202138944469168e-06,
"loss": 0.3819340467453003,
"step": 789
},
{
"epoch": 1.0821917808219177,
"grad_norm": 0.13437484204769135,
"learning_rate": 9.18002526754592e-06,
"loss": 0.32509416341781616,
"step": 790
},
{
"epoch": 1.0835616438356164,
"grad_norm": 0.14876477420330048,
"learning_rate": 9.157915626830587e-06,
"loss": 0.3575145900249481,
"step": 791
},
{
"epoch": 1.084931506849315,
"grad_norm": 0.14783765375614166,
"learning_rate": 9.135810131154707e-06,
"loss": 0.33961328864097595,
"step": 792
},
{
"epoch": 1.0863013698630137,
"grad_norm": 0.16673436760902405,
"learning_rate": 9.113708889329413e-06,
"loss": 0.3501148223876953,
"step": 793
},
{
"epoch": 1.0876712328767124,
"grad_norm": 0.14033997058868408,
"learning_rate": 9.09161201014489e-06,
"loss": 0.33308887481689453,
"step": 794
},
{
"epoch": 1.0890410958904109,
"grad_norm": 0.16320814192295074,
"learning_rate": 9.069519602369856e-06,
"loss": 0.3820383548736572,
"step": 795
},
{
"epoch": 1.0904109589041096,
"grad_norm": 0.14066532254219055,
"learning_rate": 9.047431774751019e-06,
"loss": 0.3586445748806,
"step": 796
},
{
"epoch": 1.0917808219178082,
"grad_norm": 0.14621664583683014,
"learning_rate": 9.025348636012537e-06,
"loss": 0.36413002014160156,
"step": 797
},
{
"epoch": 1.093150684931507,
"grad_norm": 0.1572379171848297,
"learning_rate": 9.003270294855491e-06,
"loss": 0.40424758195877075,
"step": 798
},
{
"epoch": 1.0945205479452054,
"grad_norm": 0.1377847045660019,
"learning_rate": 8.981196859957352e-06,
"loss": 0.3462640047073364,
"step": 799
},
{
"epoch": 1.095890410958904,
"grad_norm": 0.16429531574249268,
"learning_rate": 8.959128439971427e-06,
"loss": 0.384560227394104,
"step": 800
},
{
"epoch": 1.0972602739726027,
"grad_norm": 0.14517253637313843,
"learning_rate": 8.937065143526349e-06,
"loss": 0.3339725136756897,
"step": 801
},
{
"epoch": 1.0986301369863014,
"grad_norm": 0.16350862383842468,
"learning_rate": 8.915007079225527e-06,
"loss": 0.37034833431243896,
"step": 802
},
{
"epoch": 1.1,
"grad_norm": 0.13623249530792236,
"learning_rate": 8.892954355646616e-06,
"loss": 0.3289778232574463,
"step": 803
},
{
"epoch": 1.1013698630136985,
"grad_norm": 0.14730390906333923,
"learning_rate": 8.870907081340983e-06,
"loss": 0.3478615880012512,
"step": 804
},
{
"epoch": 1.1027397260273972,
"grad_norm": 0.14564484357833862,
"learning_rate": 8.84886536483317e-06,
"loss": 0.34111225605010986,
"step": 805
},
{
"epoch": 1.104109589041096,
"grad_norm": 0.14385566115379333,
"learning_rate": 8.826829314620367e-06,
"loss": 0.3319745361804962,
"step": 806
},
{
"epoch": 1.1054794520547946,
"grad_norm": 0.15603899955749512,
"learning_rate": 8.804799039171863e-06,
"loss": 0.3717736303806305,
"step": 807
},
{
"epoch": 1.106849315068493,
"grad_norm": 0.15575236082077026,
"learning_rate": 8.782774646928529e-06,
"loss": 0.36698096990585327,
"step": 808
},
{
"epoch": 1.1082191780821917,
"grad_norm": 0.14395324885845184,
"learning_rate": 8.760756246302277e-06,
"loss": 0.33409804105758667,
"step": 809
},
{
"epoch": 1.1095890410958904,
"grad_norm": 0.14042970538139343,
"learning_rate": 8.738743945675515e-06,
"loss": 0.3370015025138855,
"step": 810
},
{
"epoch": 1.110958904109589,
"grad_norm": 0.15795086324214935,
"learning_rate": 8.716737853400641e-06,
"loss": 0.351190447807312,
"step": 811
},
{
"epoch": 1.1123287671232878,
"grad_norm": 0.14359107613563538,
"learning_rate": 8.694738077799487e-06,
"loss": 0.3175451159477234,
"step": 812
},
{
"epoch": 1.1136986301369862,
"grad_norm": 0.1357254683971405,
"learning_rate": 8.672744727162782e-06,
"loss": 0.3094596862792969,
"step": 813
},
{
"epoch": 1.115068493150685,
"grad_norm": 0.12998807430267334,
"learning_rate": 8.650757909749643e-06,
"loss": 0.2830435633659363,
"step": 814
},
{
"epoch": 1.1164383561643836,
"grad_norm": 0.15612691640853882,
"learning_rate": 8.628777733787024e-06,
"loss": 0.30347293615341187,
"step": 815
},
{
"epoch": 1.1178082191780823,
"grad_norm": 0.16248051822185516,
"learning_rate": 8.606804307469182e-06,
"loss": 0.37846338748931885,
"step": 816
},
{
"epoch": 1.1191780821917807,
"grad_norm": 0.16550865769386292,
"learning_rate": 8.584837738957155e-06,
"loss": 0.35207441449165344,
"step": 817
},
{
"epoch": 1.1205479452054794,
"grad_norm": 0.15966041386127472,
"learning_rate": 8.562878136378226e-06,
"loss": 0.3493497967720032,
"step": 818
},
{
"epoch": 1.121917808219178,
"grad_norm": 0.15011580288410187,
"learning_rate": 8.540925607825385e-06,
"loss": 0.3301745355129242,
"step": 819
},
{
"epoch": 1.1232876712328768,
"grad_norm": 0.13884669542312622,
"learning_rate": 8.518980261356801e-06,
"loss": 0.320754736661911,
"step": 820
},
{
"epoch": 1.1246575342465754,
"grad_norm": 0.14452756941318512,
"learning_rate": 8.497042204995299e-06,
"loss": 0.32553020119667053,
"step": 821
},
{
"epoch": 1.126027397260274,
"grad_norm": 0.15732166171073914,
"learning_rate": 8.475111546727802e-06,
"loss": 0.36075225472450256,
"step": 822
},
{
"epoch": 1.1273972602739726,
"grad_norm": 0.15093612670898438,
"learning_rate": 8.453188394504834e-06,
"loss": 0.33325666189193726,
"step": 823
},
{
"epoch": 1.1287671232876713,
"grad_norm": 0.1711948662996292,
"learning_rate": 8.431272856239963e-06,
"loss": 0.3873833417892456,
"step": 824
},
{
"epoch": 1.13013698630137,
"grad_norm": 0.16041693091392517,
"learning_rate": 8.409365039809282e-06,
"loss": 0.34856122732162476,
"step": 825
},
{
"epoch": 1.1315068493150684,
"grad_norm": 0.16206581890583038,
"learning_rate": 8.38746505305087e-06,
"loss": 0.33429962396621704,
"step": 826
},
{
"epoch": 1.132876712328767,
"grad_norm": 0.14442212879657745,
"learning_rate": 8.36557300376427e-06,
"loss": 0.33349648118019104,
"step": 827
},
{
"epoch": 1.1342465753424658,
"grad_norm": 0.15366658568382263,
"learning_rate": 8.343688999709953e-06,
"loss": 0.33040598034858704,
"step": 828
},
{
"epoch": 1.1356164383561644,
"grad_norm": 0.155319482088089,
"learning_rate": 8.321813148608784e-06,
"loss": 0.37407100200653076,
"step": 829
},
{
"epoch": 1.1369863013698631,
"grad_norm": 0.17221759259700775,
"learning_rate": 8.299945558141505e-06,
"loss": 0.38914164900779724,
"step": 830
},
{
"epoch": 1.1383561643835616,
"grad_norm": 0.15503522753715515,
"learning_rate": 8.278086335948191e-06,
"loss": 0.3509390652179718,
"step": 831
},
{
"epoch": 1.1397260273972603,
"grad_norm": 0.15516449511051178,
"learning_rate": 8.25623558962772e-06,
"loss": 0.3715156316757202,
"step": 832
},
{
"epoch": 1.141095890410959,
"grad_norm": 0.16078069806098938,
"learning_rate": 8.234393426737258e-06,
"loss": 0.3550564646720886,
"step": 833
},
{
"epoch": 1.1424657534246576,
"grad_norm": 0.14807146787643433,
"learning_rate": 8.212559954791718e-06,
"loss": 0.37664794921875,
"step": 834
},
{
"epoch": 1.143835616438356,
"grad_norm": 0.1527872234582901,
"learning_rate": 8.190735281263228e-06,
"loss": 0.3674893379211426,
"step": 835
},
{
"epoch": 1.1452054794520548,
"grad_norm": 0.1597796231508255,
"learning_rate": 8.168919513580614e-06,
"loss": 0.3586568236351013,
"step": 836
},
{
"epoch": 1.1465753424657534,
"grad_norm": 0.1594488024711609,
"learning_rate": 8.147112759128859e-06,
"loss": 0.3557746708393097,
"step": 837
},
{
"epoch": 1.1479452054794521,
"grad_norm": 0.15297798812389374,
"learning_rate": 8.125315125248581e-06,
"loss": 0.36501336097717285,
"step": 838
},
{
"epoch": 1.1493150684931508,
"grad_norm": 0.16249702870845795,
"learning_rate": 8.103526719235505e-06,
"loss": 0.35979723930358887,
"step": 839
},
{
"epoch": 1.1506849315068493,
"grad_norm": 0.1662139743566513,
"learning_rate": 8.081747648339939e-06,
"loss": 0.3528470993041992,
"step": 840
},
{
"epoch": 1.152054794520548,
"grad_norm": 0.14456775784492493,
"learning_rate": 8.059978019766219e-06,
"loss": 0.35898956656455994,
"step": 841
},
{
"epoch": 1.1534246575342466,
"grad_norm": 0.14523056149482727,
"learning_rate": 8.038217940672225e-06,
"loss": 0.33172839879989624,
"step": 842
},
{
"epoch": 1.1547945205479453,
"grad_norm": 0.15215754508972168,
"learning_rate": 8.01646751816882e-06,
"loss": 0.3502211570739746,
"step": 843
},
{
"epoch": 1.1561643835616437,
"grad_norm": 0.1553926318883896,
"learning_rate": 7.994726859319348e-06,
"loss": 0.3312770128250122,
"step": 844
},
{
"epoch": 1.1575342465753424,
"grad_norm": 0.15566055476665497,
"learning_rate": 7.972996071139065e-06,
"loss": 0.33690088987350464,
"step": 845
},
{
"epoch": 1.158904109589041,
"grad_norm": 0.1474417895078659,
"learning_rate": 7.951275260594666e-06,
"loss": 0.37008821964263916,
"step": 846
},
{
"epoch": 1.1602739726027398,
"grad_norm": 0.15255552530288696,
"learning_rate": 7.929564534603722e-06,
"loss": 0.34485387802124023,
"step": 847
},
{
"epoch": 1.1616438356164385,
"grad_norm": 0.14926794171333313,
"learning_rate": 7.907864000034163e-06,
"loss": 0.3528376817703247,
"step": 848
},
{
"epoch": 1.163013698630137,
"grad_norm": 0.16398587822914124,
"learning_rate": 7.886173763703757e-06,
"loss": 0.37411195039749146,
"step": 849
},
{
"epoch": 1.1643835616438356,
"grad_norm": 0.16197367012500763,
"learning_rate": 7.86449393237958e-06,
"loss": 0.3964480757713318,
"step": 850
},
{
"epoch": 1.1657534246575343,
"grad_norm": 0.15196989476680756,
"learning_rate": 7.842824612777488e-06,
"loss": 0.3185681402683258,
"step": 851
},
{
"epoch": 1.167123287671233,
"grad_norm": 0.1505012810230255,
"learning_rate": 7.821165911561596e-06,
"loss": 0.34386467933654785,
"step": 852
},
{
"epoch": 1.1684931506849314,
"grad_norm": 0.148654505610466,
"learning_rate": 7.799517935343757e-06,
"loss": 0.34772348403930664,
"step": 853
},
{
"epoch": 1.16986301369863,
"grad_norm": 0.14560720324516296,
"learning_rate": 7.777880790683017e-06,
"loss": 0.30840951204299927,
"step": 854
},
{
"epoch": 1.1712328767123288,
"grad_norm": 0.15610314905643463,
"learning_rate": 7.756254584085121e-06,
"loss": 0.36433684825897217,
"step": 855
},
{
"epoch": 1.1726027397260275,
"grad_norm": 0.17386049032211304,
"learning_rate": 7.734639422001969e-06,
"loss": 0.38339361548423767,
"step": 856
},
{
"epoch": 1.1739726027397261,
"grad_norm": 0.1549619436264038,
"learning_rate": 7.713035410831086e-06,
"loss": 0.363090455532074,
"step": 857
},
{
"epoch": 1.1753424657534246,
"grad_norm": 0.1501813679933548,
"learning_rate": 7.691442656915122e-06,
"loss": 0.3459245562553406,
"step": 858
},
{
"epoch": 1.1767123287671233,
"grad_norm": 0.15387208759784698,
"learning_rate": 7.669861266541308e-06,
"loss": 0.3381690979003906,
"step": 859
},
{
"epoch": 1.178082191780822,
"grad_norm": 0.15518079698085785,
"learning_rate": 7.648291345940939e-06,
"loss": 0.3730364143848419,
"step": 860
},
{
"epoch": 1.1794520547945206,
"grad_norm": 0.14609061181545258,
"learning_rate": 7.626733001288852e-06,
"loss": 0.34509041905403137,
"step": 861
},
{
"epoch": 1.180821917808219,
"grad_norm": 0.14559654891490936,
"learning_rate": 7.605186338702906e-06,
"loss": 0.331429123878479,
"step": 862
},
{
"epoch": 1.1821917808219178,
"grad_norm": 0.15569762885570526,
"learning_rate": 7.583651464243458e-06,
"loss": 0.3395143151283264,
"step": 863
},
{
"epoch": 1.1835616438356165,
"grad_norm": 0.15405142307281494,
"learning_rate": 7.56212848391283e-06,
"loss": 0.3272190988063812,
"step": 864
},
{
"epoch": 1.1849315068493151,
"grad_norm": 0.14201219379901886,
"learning_rate": 7.540617503654804e-06,
"loss": 0.31572702527046204,
"step": 865
},
{
"epoch": 1.1863013698630138,
"grad_norm": 0.15996485948562622,
"learning_rate": 7.519118629354097e-06,
"loss": 0.3805091083049774,
"step": 866
},
{
"epoch": 1.1876712328767123,
"grad_norm": 0.1474556028842926,
"learning_rate": 7.497631966835828e-06,
"loss": 0.3384234607219696,
"step": 867
},
{
"epoch": 1.189041095890411,
"grad_norm": 0.1635613590478897,
"learning_rate": 7.476157621865009e-06,
"loss": 0.36581939458847046,
"step": 868
},
{
"epoch": 1.1904109589041096,
"grad_norm": 0.14194610714912415,
"learning_rate": 7.454695700146024e-06,
"loss": 0.3391876220703125,
"step": 869
},
{
"epoch": 1.191780821917808,
"grad_norm": 0.15895456075668335,
"learning_rate": 7.433246307322099e-06,
"loss": 0.35695016384124756,
"step": 870
},
{
"epoch": 1.1931506849315068,
"grad_norm": 0.1456545740365982,
"learning_rate": 7.411809548974792e-06,
"loss": 0.3398365378379822,
"step": 871
},
{
"epoch": 1.1945205479452055,
"grad_norm": 0.1395866423845291,
"learning_rate": 7.390385530623474e-06,
"loss": 0.3274722993373871,
"step": 872
},
{
"epoch": 1.1958904109589041,
"grad_norm": 0.1672808676958084,
"learning_rate": 7.368974357724789e-06,
"loss": 0.35032933950424194,
"step": 873
},
{
"epoch": 1.1972602739726028,
"grad_norm": 0.15513154864311218,
"learning_rate": 7.347576135672169e-06,
"loss": 0.34887978434562683,
"step": 874
},
{
"epoch": 1.1986301369863013,
"grad_norm": 0.16489985585212708,
"learning_rate": 7.3261909697952945e-06,
"loss": 0.33836135268211365,
"step": 875
},
{
"epoch": 1.2,
"grad_norm": 0.14997155964374542,
"learning_rate": 7.304818965359567e-06,
"loss": 0.36518996953964233,
"step": 876
},
{
"epoch": 1.2013698630136986,
"grad_norm": 0.15376335382461548,
"learning_rate": 7.283460227565614e-06,
"loss": 0.3639791011810303,
"step": 877
},
{
"epoch": 1.2027397260273973,
"grad_norm": 0.14418481290340424,
"learning_rate": 7.2621148615487595e-06,
"loss": 0.33180147409439087,
"step": 878
},
{
"epoch": 1.2041095890410958,
"grad_norm": 0.1524600386619568,
"learning_rate": 7.2407829723784965e-06,
"loss": 0.36285215616226196,
"step": 879
},
{
"epoch": 1.2054794520547945,
"grad_norm": 0.15113642811775208,
"learning_rate": 7.219464665057992e-06,
"loss": 0.3481651544570923,
"step": 880
},
{
"epoch": 1.2068493150684931,
"grad_norm": 0.1611556112766266,
"learning_rate": 7.198160044523555e-06,
"loss": 0.3910377025604248,
"step": 881
},
{
"epoch": 1.2082191780821918,
"grad_norm": 0.1525913029909134,
"learning_rate": 7.1768692156441225e-06,
"loss": 0.3326098918914795,
"step": 882
},
{
"epoch": 1.2095890410958905,
"grad_norm": 0.15305975079536438,
"learning_rate": 7.155592283220738e-06,
"loss": 0.3772580027580261,
"step": 883
},
{
"epoch": 1.210958904109589,
"grad_norm": 0.1488005518913269,
"learning_rate": 7.134329351986051e-06,
"loss": 0.34765511751174927,
"step": 884
},
{
"epoch": 1.2123287671232876,
"grad_norm": 0.1447150856256485,
"learning_rate": 7.113080526603793e-06,
"loss": 0.3677937388420105,
"step": 885
},
{
"epoch": 1.2136986301369863,
"grad_norm": 0.13866224884986877,
"learning_rate": 7.091845911668247e-06,
"loss": 0.3294437527656555,
"step": 886
},
{
"epoch": 1.215068493150685,
"grad_norm": 0.1538870632648468,
"learning_rate": 7.070625611703762e-06,
"loss": 0.3603002429008484,
"step": 887
},
{
"epoch": 1.2164383561643834,
"grad_norm": 0.16252990067005157,
"learning_rate": 7.049419731164221e-06,
"loss": 0.3410184681415558,
"step": 888
},
{
"epoch": 1.2178082191780821,
"grad_norm": 0.16610203683376312,
"learning_rate": 7.028228374432523e-06,
"loss": 0.38332611322402954,
"step": 889
},
{
"epoch": 1.2191780821917808,
"grad_norm": 0.1552305817604065,
"learning_rate": 7.0070516458200796e-06,
"loss": 0.35567736625671387,
"step": 890
},
{
"epoch": 1.2205479452054795,
"grad_norm": 0.14695027470588684,
"learning_rate": 6.9858896495663046e-06,
"loss": 0.3290688097476959,
"step": 891
},
{
"epoch": 1.2219178082191782,
"grad_norm": 0.14867903292179108,
"learning_rate": 6.964742489838076e-06,
"loss": 0.33797699213027954,
"step": 892
},
{
"epoch": 1.2232876712328766,
"grad_norm": 0.15528333187103271,
"learning_rate": 6.943610270729261e-06,
"loss": 0.3499218225479126,
"step": 893
},
{
"epoch": 1.2246575342465753,
"grad_norm": 0.16797314584255219,
"learning_rate": 6.922493096260174e-06,
"loss": 0.3884564936161041,
"step": 894
},
{
"epoch": 1.226027397260274,
"grad_norm": 0.17318569123744965,
"learning_rate": 6.901391070377069e-06,
"loss": 0.3762510120868683,
"step": 895
},
{
"epoch": 1.2273972602739727,
"grad_norm": 0.16234679520130157,
"learning_rate": 6.8803042969516436e-06,
"loss": 0.34052926301956177,
"step": 896
},
{
"epoch": 1.2287671232876711,
"grad_norm": 0.1444152593612671,
"learning_rate": 6.859232879780515e-06,
"loss": 0.32151156663894653,
"step": 897
},
{
"epoch": 1.2301369863013698,
"grad_norm": 0.1440250128507614,
"learning_rate": 6.838176922584705e-06,
"loss": 0.3263610005378723,
"step": 898
},
{
"epoch": 1.2315068493150685,
"grad_norm": 0.16781695187091827,
"learning_rate": 6.817136529009142e-06,
"loss": 0.3730999827384949,
"step": 899
},
{
"epoch": 1.2328767123287672,
"grad_norm": 0.1670355349779129,
"learning_rate": 6.796111802622148e-06,
"loss": 0.3725475072860718,
"step": 900
},
{
"epoch": 1.2342465753424658,
"grad_norm": 0.15051984786987305,
"learning_rate": 6.775102846914912e-06,
"loss": 0.35075828433036804,
"step": 901
},
{
"epoch": 1.2356164383561643,
"grad_norm": 0.1495867669582367,
"learning_rate": 6.754109765301009e-06,
"loss": 0.3146056830883026,
"step": 902
},
{
"epoch": 1.236986301369863,
"grad_norm": 0.1572950780391693,
"learning_rate": 6.73313266111587e-06,
"loss": 0.3572334945201874,
"step": 903
},
{
"epoch": 1.2383561643835617,
"grad_norm": 0.158525750041008,
"learning_rate": 6.71217163761628e-06,
"loss": 0.3286224603652954,
"step": 904
},
{
"epoch": 1.2397260273972603,
"grad_norm": 0.16932415962219238,
"learning_rate": 6.691226797979864e-06,
"loss": 0.35013192892074585,
"step": 905
},
{
"epoch": 1.2410958904109588,
"grad_norm": 0.13728591799736023,
"learning_rate": 6.670298245304587e-06,
"loss": 0.33664125204086304,
"step": 906
},
{
"epoch": 1.2424657534246575,
"grad_norm": 0.14894449710845947,
"learning_rate": 6.649386082608256e-06,
"loss": 0.35586464405059814,
"step": 907
},
{
"epoch": 1.2438356164383562,
"grad_norm": 0.16365444660186768,
"learning_rate": 6.62849041282798e-06,
"loss": 0.34779882431030273,
"step": 908
},
{
"epoch": 1.2452054794520548,
"grad_norm": 0.15779751539230347,
"learning_rate": 6.607611338819697e-06,
"loss": 0.34946244955062866,
"step": 909
},
{
"epoch": 1.2465753424657535,
"grad_norm": 0.1616879105567932,
"learning_rate": 6.586748963357655e-06,
"loss": 0.33573225140571594,
"step": 910
},
{
"epoch": 1.247945205479452,
"grad_norm": 0.1580272763967514,
"learning_rate": 6.5659033891338986e-06,
"loss": 0.3130621314048767,
"step": 911
},
{
"epoch": 1.2493150684931507,
"grad_norm": 0.14677324891090393,
"learning_rate": 6.5450747187577745e-06,
"loss": 0.341886043548584,
"step": 912
},
{
"epoch": 1.2506849315068493,
"grad_norm": 0.15481777489185333,
"learning_rate": 6.52426305475543e-06,
"loss": 0.3452962636947632,
"step": 913
},
{
"epoch": 1.252054794520548,
"grad_norm": 0.14472950994968414,
"learning_rate": 6.503468499569281e-06,
"loss": 0.32212120294570923,
"step": 914
},
{
"epoch": 1.2534246575342465,
"grad_norm": 0.16282163560390472,
"learning_rate": 6.48269115555755e-06,
"loss": 0.35730376839637756,
"step": 915
},
{
"epoch": 1.2547945205479452,
"grad_norm": 0.16933037340641022,
"learning_rate": 6.4619311249937346e-06,
"loss": 0.3709147274494171,
"step": 916
},
{
"epoch": 1.2561643835616438,
"grad_norm": 0.151408389210701,
"learning_rate": 6.441188510066092e-06,
"loss": 0.31271132826805115,
"step": 917
},
{
"epoch": 1.2575342465753425,
"grad_norm": 0.14706265926361084,
"learning_rate": 6.420463412877176e-06,
"loss": 0.3401098847389221,
"step": 918
},
{
"epoch": 1.2589041095890412,
"grad_norm": 0.15168197453022003,
"learning_rate": 6.3997559354433015e-06,
"loss": 0.32422471046447754,
"step": 919
},
{
"epoch": 1.2602739726027397,
"grad_norm": 0.15922562777996063,
"learning_rate": 6.379066179694051e-06,
"loss": 0.3759784400463104,
"step": 920
},
{
"epoch": 1.2616438356164383,
"grad_norm": 0.14806362986564636,
"learning_rate": 6.358394247471779e-06,
"loss": 0.33780619502067566,
"step": 921
},
{
"epoch": 1.263013698630137,
"grad_norm": 0.1651773303747177,
"learning_rate": 6.337740240531105e-06,
"loss": 0.3796270489692688,
"step": 922
},
{
"epoch": 1.2643835616438357,
"grad_norm": 0.14925968647003174,
"learning_rate": 6.317104260538415e-06,
"loss": 0.3102302551269531,
"step": 923
},
{
"epoch": 1.2657534246575342,
"grad_norm": 0.1527480185031891,
"learning_rate": 6.296486409071354e-06,
"loss": 0.3240545392036438,
"step": 924
},
{
"epoch": 1.2671232876712328,
"grad_norm": 0.1576816886663437,
"learning_rate": 6.275886787618339e-06,
"loss": 0.35163766145706177,
"step": 925
},
{
"epoch": 1.2684931506849315,
"grad_norm": 0.14566506445407867,
"learning_rate": 6.255305497578055e-06,
"loss": 0.346854031085968,
"step": 926
},
{
"epoch": 1.2698630136986302,
"grad_norm": 0.14731493592262268,
"learning_rate": 6.234742640258938e-06,
"loss": 0.3132811188697815,
"step": 927
},
{
"epoch": 1.2712328767123289,
"grad_norm": 0.16381718218326569,
"learning_rate": 6.214198316878707e-06,
"loss": 0.36311376094818115,
"step": 928
},
{
"epoch": 1.2726027397260273,
"grad_norm": 0.14275497198104858,
"learning_rate": 6.193672628563845e-06,
"loss": 0.356595903635025,
"step": 929
},
{
"epoch": 1.273972602739726,
"grad_norm": 0.13727188110351562,
"learning_rate": 6.173165676349103e-06,
"loss": 0.33459600806236267,
"step": 930
},
{
"epoch": 1.2753424657534247,
"grad_norm": 0.16428661346435547,
"learning_rate": 6.152677561177014e-06,
"loss": 0.36485645174980164,
"step": 931
},
{
"epoch": 1.2767123287671232,
"grad_norm": 0.15510152280330658,
"learning_rate": 6.132208383897382e-06,
"loss": 0.35984355211257935,
"step": 932
},
{
"epoch": 1.2780821917808218,
"grad_norm": 0.16664175689220428,
"learning_rate": 6.111758245266795e-06,
"loss": 0.3912050127983093,
"step": 933
},
{
"epoch": 1.2794520547945205,
"grad_norm": 0.1590886116027832,
"learning_rate": 6.091327245948122e-06,
"loss": 0.363959938287735,
"step": 934
},
{
"epoch": 1.2808219178082192,
"grad_norm": 0.1603550761938095,
"learning_rate": 6.070915486510032e-06,
"loss": 0.321783185005188,
"step": 935
},
{
"epoch": 1.2821917808219179,
"grad_norm": 0.14961084723472595,
"learning_rate": 6.05052306742647e-06,
"loss": 0.3319593369960785,
"step": 936
},
{
"epoch": 1.2835616438356166,
"grad_norm": 0.15170615911483765,
"learning_rate": 6.030150089076199e-06,
"loss": 0.3487747013568878,
"step": 937
},
{
"epoch": 1.284931506849315,
"grad_norm": 0.1593279242515564,
"learning_rate": 6.0097966517422815e-06,
"loss": 0.3525486886501312,
"step": 938
},
{
"epoch": 1.2863013698630137,
"grad_norm": 0.15932054817676544,
"learning_rate": 5.989462855611585e-06,
"loss": 0.34238719940185547,
"step": 939
},
{
"epoch": 1.2876712328767124,
"grad_norm": 0.16557934880256653,
"learning_rate": 5.969148800774307e-06,
"loss": 0.34649398922920227,
"step": 940
},
{
"epoch": 1.2890410958904108,
"grad_norm": 0.15343864262104034,
"learning_rate": 5.948854587223465e-06,
"loss": 0.3266071677207947,
"step": 941
},
{
"epoch": 1.2904109589041095,
"grad_norm": 0.1611465960741043,
"learning_rate": 5.9285803148544155e-06,
"loss": 0.39504730701446533,
"step": 942
},
{
"epoch": 1.2917808219178082,
"grad_norm": 0.1633341759443283,
"learning_rate": 5.908326083464348e-06,
"loss": 0.3515317440032959,
"step": 943
},
{
"epoch": 1.2931506849315069,
"grad_norm": 0.16506315767765045,
"learning_rate": 5.888091992751813e-06,
"loss": 0.36051037907600403,
"step": 944
},
{
"epoch": 1.2945205479452055,
"grad_norm": 0.15604349970817566,
"learning_rate": 5.867878142316221e-06,
"loss": 0.33453309535980225,
"step": 945
},
{
"epoch": 1.2958904109589042,
"grad_norm": 0.15903857350349426,
"learning_rate": 5.847684631657342e-06,
"loss": 0.35599592328071594,
"step": 946
},
{
"epoch": 1.2972602739726027,
"grad_norm": 0.15683969855308533,
"learning_rate": 5.827511560174835e-06,
"loss": 0.3563147783279419,
"step": 947
},
{
"epoch": 1.2986301369863014,
"grad_norm": 0.14779183268547058,
"learning_rate": 5.807359027167753e-06,
"loss": 0.36597883701324463,
"step": 948
},
{
"epoch": 1.3,
"grad_norm": 0.15794481337070465,
"learning_rate": 5.787227131834043e-06,
"loss": 0.3456972539424896,
"step": 949
},
{
"epoch": 1.3013698630136985,
"grad_norm": 0.1600772589445114,
"learning_rate": 5.767115973270074e-06,
"loss": 0.34236449003219604,
"step": 950
},
{
"epoch": 1.3027397260273972,
"grad_norm": 0.15741156041622162,
"learning_rate": 5.747025650470135e-06,
"loss": 0.36304205656051636,
"step": 951
},
{
"epoch": 1.3041095890410959,
"grad_norm": 0.14944733679294586,
"learning_rate": 5.7269562623259556e-06,
"loss": 0.31411996483802795,
"step": 952
},
{
"epoch": 1.3054794520547945,
"grad_norm": 0.15955956280231476,
"learning_rate": 5.706907907626216e-06,
"loss": 0.3662833273410797,
"step": 953
},
{
"epoch": 1.3068493150684932,
"grad_norm": 0.14548800885677338,
"learning_rate": 5.68688068505607e-06,
"loss": 0.3283267021179199,
"step": 954
},
{
"epoch": 1.308219178082192,
"grad_norm": 0.15583062171936035,
"learning_rate": 5.666874693196643e-06,
"loss": 0.3598661720752716,
"step": 955
},
{
"epoch": 1.3095890410958904,
"grad_norm": 0.1524307280778885,
"learning_rate": 5.646890030524555e-06,
"loss": 0.3210477828979492,
"step": 956
},
{
"epoch": 1.310958904109589,
"grad_norm": 0.15496547520160675,
"learning_rate": 5.626926795411447e-06,
"loss": 0.3147585391998291,
"step": 957
},
{
"epoch": 1.3123287671232877,
"grad_norm": 0.1605987250804901,
"learning_rate": 5.606985086123473e-06,
"loss": 0.34943753480911255,
"step": 958
},
{
"epoch": 1.3136986301369862,
"grad_norm": 0.15094715356826782,
"learning_rate": 5.587065000820833e-06,
"loss": 0.33577096462249756,
"step": 959
},
{
"epoch": 1.3150684931506849,
"grad_norm": 0.15211349725723267,
"learning_rate": 5.567166637557293e-06,
"loss": 0.32795485854148865,
"step": 960
},
{
"epoch": 1.3164383561643835,
"grad_norm": 0.1556875854730606,
"learning_rate": 5.547290094279687e-06,
"loss": 0.34780022501945496,
"step": 961
},
{
"epoch": 1.3178082191780822,
"grad_norm": 0.15435762703418732,
"learning_rate": 5.527435468827447e-06,
"loss": 0.3325795531272888,
"step": 962
},
{
"epoch": 1.319178082191781,
"grad_norm": 0.1591223031282425,
"learning_rate": 5.507602858932113e-06,
"loss": 0.3484407663345337,
"step": 963
},
{
"epoch": 1.3205479452054796,
"grad_norm": 0.1746034473180771,
"learning_rate": 5.487792362216866e-06,
"loss": 0.32427704334259033,
"step": 964
},
{
"epoch": 1.321917808219178,
"grad_norm": 0.16266120970249176,
"learning_rate": 5.46800407619603e-06,
"loss": 0.3667852282524109,
"step": 965
},
{
"epoch": 1.3232876712328767,
"grad_norm": 0.16109849512577057,
"learning_rate": 5.4482380982745985e-06,
"loss": 0.354689359664917,
"step": 966
},
{
"epoch": 1.3246575342465754,
"grad_norm": 0.15438447892665863,
"learning_rate": 5.428494525747769e-06,
"loss": 0.3435083031654358,
"step": 967
},
{
"epoch": 1.3260273972602739,
"grad_norm": 0.1704176366329193,
"learning_rate": 5.40877345580044e-06,
"loss": 0.38146907091140747,
"step": 968
},
{
"epoch": 1.3273972602739725,
"grad_norm": 0.13760878145694733,
"learning_rate": 5.38907498550674e-06,
"loss": 0.33243805170059204,
"step": 969
},
{
"epoch": 1.3287671232876712,
"grad_norm": 0.1611378937959671,
"learning_rate": 5.369399211829578e-06,
"loss": 0.36293911933898926,
"step": 970
},
{
"epoch": 1.33013698630137,
"grad_norm": 0.15081392228603363,
"learning_rate": 5.3497462316201055e-06,
"loss": 0.36365821957588196,
"step": 971
},
{
"epoch": 1.3315068493150686,
"grad_norm": 0.1435474306344986,
"learning_rate": 5.330116141617308e-06,
"loss": 0.33524560928344727,
"step": 972
},
{
"epoch": 1.3328767123287673,
"grad_norm": 0.1512869894504547,
"learning_rate": 5.310509038447492e-06,
"loss": 0.35711202025413513,
"step": 973
},
{
"epoch": 1.3342465753424657,
"grad_norm": 0.16022555530071259,
"learning_rate": 5.2909250186237935e-06,
"loss": 0.3644453287124634,
"step": 974
},
{
"epoch": 1.3356164383561644,
"grad_norm": 0.1494358479976654,
"learning_rate": 5.2713641785457504e-06,
"loss": 0.3420622944831848,
"step": 975
},
{
"epoch": 1.336986301369863,
"grad_norm": 0.14082303643226624,
"learning_rate": 5.251826614498785e-06,
"loss": 0.30518871545791626,
"step": 976
},
{
"epoch": 1.3383561643835615,
"grad_norm": 0.13758394122123718,
"learning_rate": 5.23231242265375e-06,
"loss": 0.31102830171585083,
"step": 977
},
{
"epoch": 1.3397260273972602,
"grad_norm": 0.15121538937091827,
"learning_rate": 5.212821699066459e-06,
"loss": 0.36003631353378296,
"step": 978
},
{
"epoch": 1.341095890410959,
"grad_norm": 0.15851706266403198,
"learning_rate": 5.1933545396772e-06,
"loss": 0.35232266783714294,
"step": 979
},
{
"epoch": 1.3424657534246576,
"grad_norm": 0.1462315171957016,
"learning_rate": 5.173911040310266e-06,
"loss": 0.31785115599632263,
"step": 980
},
{
"epoch": 1.3438356164383563,
"grad_norm": 0.16444040834903717,
"learning_rate": 5.1544912966735e-06,
"loss": 0.387692928314209,
"step": 981
},
{
"epoch": 1.345205479452055,
"grad_norm": 0.14166902005672455,
"learning_rate": 5.135095404357793e-06,
"loss": 0.2919553220272064,
"step": 982
},
{
"epoch": 1.3465753424657534,
"grad_norm": 0.149844229221344,
"learning_rate": 5.115723458836652e-06,
"loss": 0.3367069363594055,
"step": 983
},
{
"epoch": 1.347945205479452,
"grad_norm": 0.1612461507320404,
"learning_rate": 5.0963755554656856e-06,
"loss": 0.35543060302734375,
"step": 984
},
{
"epoch": 1.3493150684931507,
"grad_norm": 0.1738181859254837,
"learning_rate": 5.0770517894821815e-06,
"loss": 0.3604719936847687,
"step": 985
},
{
"epoch": 1.3506849315068492,
"grad_norm": 0.17190732061862946,
"learning_rate": 5.057752256004597e-06,
"loss": 0.3398797810077667,
"step": 986
},
{
"epoch": 1.3520547945205479,
"grad_norm": 0.1579580157995224,
"learning_rate": 5.0384770500321175e-06,
"loss": 0.30709680914878845,
"step": 987
},
{
"epoch": 1.3534246575342466,
"grad_norm": 0.1556752473115921,
"learning_rate": 5.01922626644418e-06,
"loss": 0.33867329359054565,
"step": 988
},
{
"epoch": 1.3547945205479452,
"grad_norm": 0.15582862496376038,
"learning_rate": 5.000000000000003e-06,
"loss": 0.3532813787460327,
"step": 989
},
{
"epoch": 1.356164383561644,
"grad_norm": 0.16169165074825287,
"learning_rate": 4.980798345338117e-06,
"loss": 0.33946484327316284,
"step": 990
},
{
"epoch": 1.3575342465753424,
"grad_norm": 0.15696866810321808,
"learning_rate": 4.961621396975919e-06,
"loss": 0.34943240880966187,
"step": 991
},
{
"epoch": 1.358904109589041,
"grad_norm": 0.16400301456451416,
"learning_rate": 4.942469249309182e-06,
"loss": 0.3292185664176941,
"step": 992
},
{
"epoch": 1.3602739726027397,
"grad_norm": 0.16433073580265045,
"learning_rate": 4.923341996611604e-06,
"loss": 0.3521850109100342,
"step": 993
},
{
"epoch": 1.3616438356164384,
"grad_norm": 0.14444899559020996,
"learning_rate": 4.904239733034335e-06,
"loss": 0.30568253993988037,
"step": 994
},
{
"epoch": 1.3630136986301369,
"grad_norm": 0.15330424904823303,
"learning_rate": 4.8851625526055345e-06,
"loss": 0.33213597536087036,
"step": 995
},
{
"epoch": 1.3643835616438356,
"grad_norm": 0.16560016572475433,
"learning_rate": 4.866110549229881e-06,
"loss": 0.35149186849594116,
"step": 996
},
{
"epoch": 1.3657534246575342,
"grad_norm": 0.16279006004333496,
"learning_rate": 4.847083816688123e-06,
"loss": 0.31754031777381897,
"step": 997
},
{
"epoch": 1.367123287671233,
"grad_norm": 0.16460226476192474,
"learning_rate": 4.828082448636626e-06,
"loss": 0.33603787422180176,
"step": 998
},
{
"epoch": 1.3684931506849316,
"grad_norm": 0.15313956141471863,
"learning_rate": 4.809106538606896e-06,
"loss": 0.3381017744541168,
"step": 999
},
{
"epoch": 1.36986301369863,
"grad_norm": 0.15496991574764252,
"learning_rate": 4.790156180005119e-06,
"loss": 0.3478963375091553,
"step": 1000
},
{
"epoch": 1.3712328767123287,
"grad_norm": 0.15212573111057281,
"learning_rate": 4.771231466111725e-06,
"loss": 0.368503212928772,
"step": 1001
},
{
"epoch": 1.3726027397260274,
"grad_norm": 0.15523137152194977,
"learning_rate": 4.7523324900808986e-06,
"loss": 0.35156184434890747,
"step": 1002
},
{
"epoch": 1.373972602739726,
"grad_norm": 0.16248948872089386,
"learning_rate": 4.73345934494013e-06,
"loss": 0.35914623737335205,
"step": 1003
},
{
"epoch": 1.3753424657534246,
"grad_norm": 0.16289544105529785,
"learning_rate": 4.714612123589777e-06,
"loss": 0.35127484798431396,
"step": 1004
},
{
"epoch": 1.3767123287671232,
"grad_norm": 0.15264225006103516,
"learning_rate": 4.695790918802577e-06,
"loss": 0.36287033557891846,
"step": 1005
},
{
"epoch": 1.378082191780822,
"grad_norm": 0.150607630610466,
"learning_rate": 4.676995823223207e-06,
"loss": 0.3071051239967346,
"step": 1006
},
{
"epoch": 1.3794520547945206,
"grad_norm": 0.15527276694774628,
"learning_rate": 4.658226929367826e-06,
"loss": 0.3501741290092468,
"step": 1007
},
{
"epoch": 1.3808219178082193,
"grad_norm": 0.15097099542617798,
"learning_rate": 4.639484329623627e-06,
"loss": 0.33700865507125854,
"step": 1008
},
{
"epoch": 1.3821917808219177,
"grad_norm": 0.1395462155342102,
"learning_rate": 4.620768116248362e-06,
"loss": 0.3207249641418457,
"step": 1009
},
{
"epoch": 1.3835616438356164,
"grad_norm": 0.16464103758335114,
"learning_rate": 4.6020783813699024e-06,
"loss": 0.35011178255081177,
"step": 1010
},
{
"epoch": 1.384931506849315,
"grad_norm": 0.1423029750585556,
"learning_rate": 4.583415216985791e-06,
"loss": 0.3034648895263672,
"step": 1011
},
{
"epoch": 1.3863013698630138,
"grad_norm": 0.1641887128353119,
"learning_rate": 4.564778714962773e-06,
"loss": 0.3178097605705261,
"step": 1012
},
{
"epoch": 1.3876712328767122,
"grad_norm": 0.16252465546131134,
"learning_rate": 4.546168967036352e-06,
"loss": 0.3473752439022064,
"step": 1013
},
{
"epoch": 1.389041095890411,
"grad_norm": 0.15077027678489685,
"learning_rate": 4.5275860648103496e-06,
"loss": 0.3326604962348938,
"step": 1014
},
{
"epoch": 1.3904109589041096,
"grad_norm": 0.16695444285869598,
"learning_rate": 4.50903009975642e-06,
"loss": 0.3639211654663086,
"step": 1015
},
{
"epoch": 1.3917808219178083,
"grad_norm": 0.16211333870887756,
"learning_rate": 4.490501163213652e-06,
"loss": 0.365367591381073,
"step": 1016
},
{
"epoch": 1.393150684931507,
"grad_norm": 0.15747344493865967,
"learning_rate": 4.4719993463880695e-06,
"loss": 0.35200953483581543,
"step": 1017
},
{
"epoch": 1.3945205479452054,
"grad_norm": 0.15495389699935913,
"learning_rate": 4.453524740352207e-06,
"loss": 0.34566500782966614,
"step": 1018
},
{
"epoch": 1.395890410958904,
"grad_norm": 0.16389113664627075,
"learning_rate": 4.435077436044667e-06,
"loss": 0.35072052478790283,
"step": 1019
},
{
"epoch": 1.3972602739726028,
"grad_norm": 0.15782539546489716,
"learning_rate": 4.416657524269652e-06,
"loss": 0.31635719537734985,
"step": 1020
},
{
"epoch": 1.3986301369863012,
"grad_norm": 0.166508287191391,
"learning_rate": 4.398265095696539e-06,
"loss": 0.3655712902545929,
"step": 1021
},
{
"epoch": 1.4,
"grad_norm": 0.16588471829891205,
"learning_rate": 4.379900240859416e-06,
"loss": 0.34234797954559326,
"step": 1022
},
{
"epoch": 1.4013698630136986,
"grad_norm": 0.16490907967090607,
"learning_rate": 4.361563050156639e-06,
"loss": 0.33568212389945984,
"step": 1023
},
{
"epoch": 1.4027397260273973,
"grad_norm": 0.14453092217445374,
"learning_rate": 4.343253613850409e-06,
"loss": 0.3196790814399719,
"step": 1024
},
{
"epoch": 1.404109589041096,
"grad_norm": 0.16436739265918732,
"learning_rate": 4.324972022066284e-06,
"loss": 0.3360899090766907,
"step": 1025
},
{
"epoch": 1.4054794520547946,
"grad_norm": 0.15080416202545166,
"learning_rate": 4.3067183647927855e-06,
"loss": 0.3406558632850647,
"step": 1026
},
{
"epoch": 1.406849315068493,
"grad_norm": 0.1640864461660385,
"learning_rate": 4.288492731880917e-06,
"loss": 0.3594125509262085,
"step": 1027
},
{
"epoch": 1.4082191780821918,
"grad_norm": 0.15736639499664307,
"learning_rate": 4.270295213043741e-06,
"loss": 0.3553820550441742,
"step": 1028
},
{
"epoch": 1.4095890410958904,
"grad_norm": 0.15510410070419312,
"learning_rate": 4.2521258978559324e-06,
"loss": 0.3318788409233093,
"step": 1029
},
{
"epoch": 1.410958904109589,
"grad_norm": 0.15620078146457672,
"learning_rate": 4.233984875753338e-06,
"loss": 0.3684927523136139,
"step": 1030
},
{
"epoch": 1.4123287671232876,
"grad_norm": 0.14800132811069489,
"learning_rate": 4.215872236032531e-06,
"loss": 0.32929253578186035,
"step": 1031
},
{
"epoch": 1.4136986301369863,
"grad_norm": 0.16534818708896637,
"learning_rate": 4.197788067850388e-06,
"loss": 0.3849608600139618,
"step": 1032
},
{
"epoch": 1.415068493150685,
"grad_norm": 0.15061511099338531,
"learning_rate": 4.179732460223626e-06,
"loss": 0.3256608545780182,
"step": 1033
},
{
"epoch": 1.4164383561643836,
"grad_norm": 0.1583685278892517,
"learning_rate": 4.161705502028382e-06,
"loss": 0.3300684094429016,
"step": 1034
},
{
"epoch": 1.4178082191780823,
"grad_norm": 0.1482117921113968,
"learning_rate": 4.143707281999767e-06,
"loss": 0.33890092372894287,
"step": 1035
},
{
"epoch": 1.4191780821917808,
"grad_norm": 0.1462375372648239,
"learning_rate": 4.125737888731444e-06,
"loss": 0.31467318534851074,
"step": 1036
},
{
"epoch": 1.4205479452054794,
"grad_norm": 0.1618596315383911,
"learning_rate": 4.107797410675166e-06,
"loss": 0.31928420066833496,
"step": 1037
},
{
"epoch": 1.4219178082191781,
"grad_norm": 0.15106843411922455,
"learning_rate": 4.0898859361403595e-06,
"loss": 0.32146143913269043,
"step": 1038
},
{
"epoch": 1.4232876712328766,
"grad_norm": 0.1492437720298767,
"learning_rate": 4.072003553293693e-06,
"loss": 0.32602545619010925,
"step": 1039
},
{
"epoch": 1.4246575342465753,
"grad_norm": 0.16238275170326233,
"learning_rate": 4.054150350158626e-06,
"loss": 0.36526453495025635,
"step": 1040
},
{
"epoch": 1.426027397260274,
"grad_norm": 0.14458993077278137,
"learning_rate": 4.036326414614985e-06,
"loss": 0.3100438714027405,
"step": 1041
},
{
"epoch": 1.4273972602739726,
"grad_norm": 0.14486831426620483,
"learning_rate": 4.0185318343985415e-06,
"loss": 0.33524399995803833,
"step": 1042
},
{
"epoch": 1.4287671232876713,
"grad_norm": 0.1474289894104004,
"learning_rate": 4.000766697100558e-06,
"loss": 0.3067096471786499,
"step": 1043
},
{
"epoch": 1.43013698630137,
"grad_norm": 0.1528017520904541,
"learning_rate": 3.983031090167368e-06,
"loss": 0.3176173269748688,
"step": 1044
},
{
"epoch": 1.4315068493150684,
"grad_norm": 0.1463933289051056,
"learning_rate": 3.9653251008999615e-06,
"loss": 0.32957181334495544,
"step": 1045
},
{
"epoch": 1.4328767123287671,
"grad_norm": 0.1470157951116562,
"learning_rate": 3.947648816453523e-06,
"loss": 0.3459767699241638,
"step": 1046
},
{
"epoch": 1.4342465753424658,
"grad_norm": 0.16025510430335999,
"learning_rate": 3.930002323837026e-06,
"loss": 0.3682284653186798,
"step": 1047
},
{
"epoch": 1.4356164383561643,
"grad_norm": 0.15199273824691772,
"learning_rate": 3.912385709912794e-06,
"loss": 0.34587031602859497,
"step": 1048
},
{
"epoch": 1.436986301369863,
"grad_norm": 0.1603536605834961,
"learning_rate": 3.894799061396086e-06,
"loss": 0.33916258811950684,
"step": 1049
},
{
"epoch": 1.4383561643835616,
"grad_norm": 0.1427288055419922,
"learning_rate": 3.877242464854654e-06,
"loss": 0.3523396849632263,
"step": 1050
},
{
"epoch": 1.4397260273972603,
"grad_norm": 0.18065136671066284,
"learning_rate": 3.85971600670832e-06,
"loss": 0.3822803497314453,
"step": 1051
},
{
"epoch": 1.441095890410959,
"grad_norm": 0.15296302735805511,
"learning_rate": 3.842219773228565e-06,
"loss": 0.33401668071746826,
"step": 1052
},
{
"epoch": 1.4424657534246577,
"grad_norm": 0.14656171202659607,
"learning_rate": 3.824753850538082e-06,
"loss": 0.33078229427337646,
"step": 1053
},
{
"epoch": 1.4438356164383561,
"grad_norm": 0.16649788618087769,
"learning_rate": 3.8073183246103672e-06,
"loss": 0.361587256193161,
"step": 1054
},
{
"epoch": 1.4452054794520548,
"grad_norm": 0.15856561064720154,
"learning_rate": 3.789913281269304e-06,
"loss": 0.35494107007980347,
"step": 1055
},
{
"epoch": 1.4465753424657535,
"grad_norm": 0.1742408275604248,
"learning_rate": 3.7725388061887056e-06,
"loss": 0.35915690660476685,
"step": 1056
},
{
"epoch": 1.447945205479452,
"grad_norm": 0.15447255969047546,
"learning_rate": 3.755194984891943e-06,
"loss": 0.32169297337532043,
"step": 1057
},
{
"epoch": 1.4493150684931506,
"grad_norm": 0.15367941558361053,
"learning_rate": 3.737881902751479e-06,
"loss": 0.34400704503059387,
"step": 1058
},
{
"epoch": 1.4506849315068493,
"grad_norm": 0.15320177376270294,
"learning_rate": 3.720599644988482e-06,
"loss": 0.3285016119480133,
"step": 1059
},
{
"epoch": 1.452054794520548,
"grad_norm": 0.15503562986850739,
"learning_rate": 3.7033482966723833e-06,
"loss": 0.3686943054199219,
"step": 1060
},
{
"epoch": 1.4534246575342467,
"grad_norm": 0.15862441062927246,
"learning_rate": 3.6861279427204634e-06,
"loss": 0.35472556948661804,
"step": 1061
},
{
"epoch": 1.4547945205479453,
"grad_norm": 0.15359939634799957,
"learning_rate": 3.6689386678974504e-06,
"loss": 0.36844033002853394,
"step": 1062
},
{
"epoch": 1.4561643835616438,
"grad_norm": 0.17726793885231018,
"learning_rate": 3.6517805568150787e-06,
"loss": 0.37768447399139404,
"step": 1063
},
{
"epoch": 1.4575342465753425,
"grad_norm": 0.15256346762180328,
"learning_rate": 3.6346536939316847e-06,
"loss": 0.32124149799346924,
"step": 1064
},
{
"epoch": 1.4589041095890412,
"grad_norm": 0.14414700865745544,
"learning_rate": 3.617558163551802e-06,
"loss": 0.3302096724510193,
"step": 1065
},
{
"epoch": 1.4602739726027396,
"grad_norm": 0.14682701230049133,
"learning_rate": 3.6004940498257114e-06,
"loss": 0.34126797318458557,
"step": 1066
},
{
"epoch": 1.4616438356164383,
"grad_norm": 0.14877453446388245,
"learning_rate": 3.5834614367490706e-06,
"loss": 0.3140515089035034,
"step": 1067
},
{
"epoch": 1.463013698630137,
"grad_norm": 0.1443350911140442,
"learning_rate": 3.5664604081624787e-06,
"loss": 0.2944907546043396,
"step": 1068
},
{
"epoch": 1.4643835616438357,
"grad_norm": 0.1500612050294876,
"learning_rate": 3.5494910477510445e-06,
"loss": 0.32540619373321533,
"step": 1069
},
{
"epoch": 1.4657534246575343,
"grad_norm": 0.1721482276916504,
"learning_rate": 3.5325534390440208e-06,
"loss": 0.3468330204486847,
"step": 1070
},
{
"epoch": 1.467123287671233,
"grad_norm": 0.1610793173313141,
"learning_rate": 3.51564766541435e-06,
"loss": 0.3564775884151459,
"step": 1071
},
{
"epoch": 1.4684931506849315,
"grad_norm": 0.15753892064094543,
"learning_rate": 3.4987738100782742e-06,
"loss": 0.365815132856369,
"step": 1072
},
{
"epoch": 1.4698630136986301,
"grad_norm": 0.14811822772026062,
"learning_rate": 3.4819319560949303e-06,
"loss": 0.3437769412994385,
"step": 1073
},
{
"epoch": 1.4712328767123288,
"grad_norm": 0.17442472279071808,
"learning_rate": 3.4651221863659236e-06,
"loss": 0.35018718242645264,
"step": 1074
},
{
"epoch": 1.4726027397260273,
"grad_norm": 0.15346552431583405,
"learning_rate": 3.4483445836349284e-06,
"loss": 0.37094342708587646,
"step": 1075
},
{
"epoch": 1.473972602739726,
"grad_norm": 0.15852433443069458,
"learning_rate": 3.4315992304872945e-06,
"loss": 0.34059029817581177,
"step": 1076
},
{
"epoch": 1.4753424657534246,
"grad_norm": 0.15421006083488464,
"learning_rate": 3.414886209349615e-06,
"loss": 0.34393078088760376,
"step": 1077
},
{
"epoch": 1.4767123287671233,
"grad_norm": 0.15218770503997803,
"learning_rate": 3.3982056024893374e-06,
"loss": 0.3340407609939575,
"step": 1078
},
{
"epoch": 1.478082191780822,
"grad_norm": 0.1413390338420868,
"learning_rate": 3.381557492014351e-06,
"loss": 0.3110498785972595,
"step": 1079
},
{
"epoch": 1.4794520547945205,
"grad_norm": 0.15303197503089905,
"learning_rate": 3.3649419598725964e-06,
"loss": 0.3134489059448242,
"step": 1080
},
{
"epoch": 1.4808219178082191,
"grad_norm": 0.1585136353969574,
"learning_rate": 3.3483590878516437e-06,
"loss": 0.33575257658958435,
"step": 1081
},
{
"epoch": 1.4821917808219178,
"grad_norm": 0.16408440470695496,
"learning_rate": 3.331808957578294e-06,
"loss": 0.353657990694046,
"step": 1082
},
{
"epoch": 1.4835616438356165,
"grad_norm": 0.14696872234344482,
"learning_rate": 3.3152916505181976e-06,
"loss": 0.30488771200180054,
"step": 1083
},
{
"epoch": 1.484931506849315,
"grad_norm": 0.1497623771429062,
"learning_rate": 3.298807247975423e-06,
"loss": 0.31840866804122925,
"step": 1084
},
{
"epoch": 1.4863013698630136,
"grad_norm": 0.16288898885250092,
"learning_rate": 3.282355831092072e-06,
"loss": 0.2969713807106018,
"step": 1085
},
{
"epoch": 1.4876712328767123,
"grad_norm": 0.15872237086296082,
"learning_rate": 3.2659374808478892e-06,
"loss": 0.351738303899765,
"step": 1086
},
{
"epoch": 1.489041095890411,
"grad_norm": 0.15343013405799866,
"learning_rate": 3.2495522780598442e-06,
"loss": 0.33581048250198364,
"step": 1087
},
{
"epoch": 1.4904109589041097,
"grad_norm": 0.15491418540477753,
"learning_rate": 3.233200303381746e-06,
"loss": 0.30864545702934265,
"step": 1088
},
{
"epoch": 1.4917808219178081,
"grad_norm": 0.17173396050930023,
"learning_rate": 3.216881637303839e-06,
"loss": 0.3643016815185547,
"step": 1089
},
{
"epoch": 1.4931506849315068,
"grad_norm": 0.1479952335357666,
"learning_rate": 3.2005963601524213e-06,
"loss": 0.33544814586639404,
"step": 1090
},
{
"epoch": 1.4945205479452055,
"grad_norm": 0.162466362118721,
"learning_rate": 3.184344552089429e-06,
"loss": 0.3340489864349365,
"step": 1091
},
{
"epoch": 1.4958904109589042,
"grad_norm": 0.14227993786334991,
"learning_rate": 3.1681262931120504e-06,
"loss": 0.29176169633865356,
"step": 1092
},
{
"epoch": 1.4972602739726026,
"grad_norm": 0.17138493061065674,
"learning_rate": 3.151941663052345e-06,
"loss": 0.3775274455547333,
"step": 1093
},
{
"epoch": 1.4986301369863013,
"grad_norm": 0.1501752734184265,
"learning_rate": 3.1357907415768264e-06,
"loss": 0.34381645917892456,
"step": 1094
},
{
"epoch": 1.5,
"grad_norm": 0.15916335582733154,
"learning_rate": 3.1196736081860855e-06,
"loss": 0.3555161952972412,
"step": 1095
},
{
"epoch": 1.5013698630136987,
"grad_norm": 0.15691401064395905,
"learning_rate": 3.1035903422144075e-06,
"loss": 0.33744144439697266,
"step": 1096
},
{
"epoch": 1.5027397260273974,
"grad_norm": 0.14738737046718597,
"learning_rate": 3.087541022829347e-06,
"loss": 0.3242756724357605,
"step": 1097
},
{
"epoch": 1.504109589041096,
"grad_norm": 0.1540118008852005,
"learning_rate": 3.0715257290313836e-06,
"loss": 0.31719303131103516,
"step": 1098
},
{
"epoch": 1.5054794520547945,
"grad_norm": 0.17718595266342163,
"learning_rate": 3.055544539653498e-06,
"loss": 0.3548113703727722,
"step": 1099
},
{
"epoch": 1.5068493150684932,
"grad_norm": 0.1463872641324997,
"learning_rate": 3.0395975333608063e-06,
"loss": 0.337434321641922,
"step": 1100
},
{
"epoch": 1.5082191780821916,
"grad_norm": 0.14244531095027924,
"learning_rate": 3.023684788650154e-06,
"loss": 0.33371126651763916,
"step": 1101
},
{
"epoch": 1.5095890410958903,
"grad_norm": 0.16751514375209808,
"learning_rate": 3.007806383849742e-06,
"loss": 0.3705282211303711,
"step": 1102
},
{
"epoch": 1.510958904109589,
"grad_norm": 0.14863881468772888,
"learning_rate": 2.9919623971187415e-06,
"loss": 0.3338697850704193,
"step": 1103
},
{
"epoch": 1.5123287671232877,
"grad_norm": 0.14219138026237488,
"learning_rate": 2.976152906446903e-06,
"loss": 0.30235278606414795,
"step": 1104
},
{
"epoch": 1.5136986301369864,
"grad_norm": 0.1501089334487915,
"learning_rate": 2.9603779896541705e-06,
"loss": 0.32276517152786255,
"step": 1105
},
{
"epoch": 1.515068493150685,
"grad_norm": 0.1589280366897583,
"learning_rate": 2.944637724390317e-06,
"loss": 0.35342103242874146,
"step": 1106
},
{
"epoch": 1.5164383561643837,
"grad_norm": 0.15434975922107697,
"learning_rate": 2.9289321881345257e-06,
"loss": 0.3250484764575958,
"step": 1107
},
{
"epoch": 1.5178082191780822,
"grad_norm": 0.1619551032781601,
"learning_rate": 2.9132614581950512e-06,
"loss": 0.3212895393371582,
"step": 1108
},
{
"epoch": 1.5191780821917809,
"grad_norm": 0.16642409563064575,
"learning_rate": 2.89762561170882e-06,
"loss": 0.3505953848361969,
"step": 1109
},
{
"epoch": 1.5205479452054793,
"grad_norm": 0.14051656424999237,
"learning_rate": 2.8820247256410272e-06,
"loss": 0.31902238726615906,
"step": 1110
},
{
"epoch": 1.521917808219178,
"grad_norm": 0.15815143287181854,
"learning_rate": 2.866458876784807e-06,
"loss": 0.33270347118377686,
"step": 1111
},
{
"epoch": 1.5232876712328767,
"grad_norm": 0.152178555727005,
"learning_rate": 2.8509281417608105e-06,
"loss": 0.33165889978408813,
"step": 1112
},
{
"epoch": 1.5246575342465754,
"grad_norm": 0.15572141110897064,
"learning_rate": 2.8354325970168483e-06,
"loss": 0.3630872964859009,
"step": 1113
},
{
"epoch": 1.526027397260274,
"grad_norm": 0.1545613557100296,
"learning_rate": 2.819972318827524e-06,
"loss": 0.3553091883659363,
"step": 1114
},
{
"epoch": 1.5273972602739727,
"grad_norm": 0.16011565923690796,
"learning_rate": 2.8045473832938332e-06,
"loss": 0.33448928594589233,
"step": 1115
},
{
"epoch": 1.5287671232876714,
"grad_norm": 0.16422204673290253,
"learning_rate": 2.7891578663428033e-06,
"loss": 0.33384954929351807,
"step": 1116
},
{
"epoch": 1.5301369863013699,
"grad_norm": 0.16461777687072754,
"learning_rate": 2.7738038437271288e-06,
"loss": 0.35072165727615356,
"step": 1117
},
{
"epoch": 1.5315068493150685,
"grad_norm": 0.17709890007972717,
"learning_rate": 2.7584853910247765e-06,
"loss": 0.359376460313797,
"step": 1118
},
{
"epoch": 1.532876712328767,
"grad_norm": 0.13950414955615997,
"learning_rate": 2.7432025836386412e-06,
"loss": 0.3026618957519531,
"step": 1119
},
{
"epoch": 1.5342465753424657,
"grad_norm": 0.15768079459667206,
"learning_rate": 2.727955496796134e-06,
"loss": 0.34044182300567627,
"step": 1120
},
{
"epoch": 1.5356164383561643,
"grad_norm": 0.1542557179927826,
"learning_rate": 2.7127442055488617e-06,
"loss": 0.3653857707977295,
"step": 1121
},
{
"epoch": 1.536986301369863,
"grad_norm": 0.1634950190782547,
"learning_rate": 2.6975687847722197e-06,
"loss": 0.35250169038772583,
"step": 1122
},
{
"epoch": 1.5383561643835617,
"grad_norm": 0.16520905494689941,
"learning_rate": 2.682429309165033e-06,
"loss": 0.3734781742095947,
"step": 1123
},
{
"epoch": 1.5397260273972604,
"grad_norm": 0.169541135430336,
"learning_rate": 2.6673258532492063e-06,
"loss": 0.3353344798088074,
"step": 1124
},
{
"epoch": 1.541095890410959,
"grad_norm": 0.15883050858974457,
"learning_rate": 2.6522584913693295e-06,
"loss": 0.37120819091796875,
"step": 1125
},
{
"epoch": 1.5424657534246575,
"grad_norm": 0.14935588836669922,
"learning_rate": 2.6372272976923242e-06,
"loss": 0.32711493968963623,
"step": 1126
},
{
"epoch": 1.5438356164383562,
"grad_norm": 0.15894056856632233,
"learning_rate": 2.6222323462070897e-06,
"loss": 0.3108447790145874,
"step": 1127
},
{
"epoch": 1.5452054794520547,
"grad_norm": 0.14813277125358582,
"learning_rate": 2.607273710724121e-06,
"loss": 0.31116819381713867,
"step": 1128
},
{
"epoch": 1.5465753424657533,
"grad_norm": 0.1491813063621521,
"learning_rate": 2.5923514648751537e-06,
"loss": 0.3273397386074066,
"step": 1129
},
{
"epoch": 1.547945205479452,
"grad_norm": 0.1643611490726471,
"learning_rate": 2.5774656821127965e-06,
"loss": 0.3463200032711029,
"step": 1130
},
{
"epoch": 1.5493150684931507,
"grad_norm": 0.15818530321121216,
"learning_rate": 2.5626164357101857e-06,
"loss": 0.36125755310058594,
"step": 1131
},
{
"epoch": 1.5506849315068494,
"grad_norm": 0.15721388161182404,
"learning_rate": 2.547803798760603e-06,
"loss": 0.32161208987236023,
"step": 1132
},
{
"epoch": 1.552054794520548,
"grad_norm": 0.1627534031867981,
"learning_rate": 2.533027844177123e-06,
"loss": 0.33864739537239075,
"step": 1133
},
{
"epoch": 1.5534246575342465,
"grad_norm": 0.15564633905887604,
"learning_rate": 2.5182886446922673e-06,
"loss": 0.3207649290561676,
"step": 1134
},
{
"epoch": 1.5547945205479452,
"grad_norm": 0.14905612170696259,
"learning_rate": 2.5035862728576266e-06,
"loss": 0.3708069920539856,
"step": 1135
},
{
"epoch": 1.5561643835616439,
"grad_norm": 0.17216545343399048,
"learning_rate": 2.488920801043514e-06,
"loss": 0.3721892237663269,
"step": 1136
},
{
"epoch": 1.5575342465753423,
"grad_norm": 0.16713808476924896,
"learning_rate": 2.4742923014386154e-06,
"loss": 0.3689742684364319,
"step": 1137
},
{
"epoch": 1.558904109589041,
"grad_norm": 0.15405254065990448,
"learning_rate": 2.4597008460496165e-06,
"loss": 0.34399300813674927,
"step": 1138
},
{
"epoch": 1.5602739726027397,
"grad_norm": 0.15538863837718964,
"learning_rate": 2.4451465067008618e-06,
"loss": 0.32398271560668945,
"step": 1139
},
{
"epoch": 1.5616438356164384,
"grad_norm": 0.16837860643863678,
"learning_rate": 2.4306293550339943e-06,
"loss": 0.35376736521720886,
"step": 1140
},
{
"epoch": 1.563013698630137,
"grad_norm": 0.16373756527900696,
"learning_rate": 2.4161494625076164e-06,
"loss": 0.3440687358379364,
"step": 1141
},
{
"epoch": 1.5643835616438357,
"grad_norm": 0.1669769138097763,
"learning_rate": 2.4017069003969174e-06,
"loss": 0.3465386927127838,
"step": 1142
},
{
"epoch": 1.5657534246575342,
"grad_norm": 0.1637868583202362,
"learning_rate": 2.387301739793333e-06,
"loss": 0.31126880645751953,
"step": 1143
},
{
"epoch": 1.5671232876712329,
"grad_norm": 0.16006502509117126,
"learning_rate": 2.3729340516042044e-06,
"loss": 0.3085182011127472,
"step": 1144
},
{
"epoch": 1.5684931506849316,
"grad_norm": 0.15065310895442963,
"learning_rate": 2.3586039065524113e-06,
"loss": 0.3250669538974762,
"step": 1145
},
{
"epoch": 1.56986301369863,
"grad_norm": 0.1626662313938141,
"learning_rate": 2.344311375176034e-06,
"loss": 0.33825933933258057,
"step": 1146
},
{
"epoch": 1.5712328767123287,
"grad_norm": 0.16704460978507996,
"learning_rate": 2.330056527828013e-06,
"loss": 0.37022721767425537,
"step": 1147
},
{
"epoch": 1.5726027397260274,
"grad_norm": 0.1803397238254547,
"learning_rate": 2.3158394346757786e-06,
"loss": 0.3628985285758972,
"step": 1148
},
{
"epoch": 1.573972602739726,
"grad_norm": 0.15835583209991455,
"learning_rate": 2.3016601657009364e-06,
"loss": 0.36137890815734863,
"step": 1149
},
{
"epoch": 1.5753424657534247,
"grad_norm": 0.15241596102714539,
"learning_rate": 2.2875187906989084e-06,
"loss": 0.34486639499664307,
"step": 1150
},
{
"epoch": 1.5767123287671234,
"grad_norm": 0.16082894802093506,
"learning_rate": 2.2734153792785706e-06,
"loss": 0.35999006032943726,
"step": 1151
},
{
"epoch": 1.5780821917808219,
"grad_norm": 0.14098484814167023,
"learning_rate": 2.259350000861952e-06,
"loss": 0.32649844884872437,
"step": 1152
},
{
"epoch": 1.5794520547945206,
"grad_norm": 0.15429721772670746,
"learning_rate": 2.245322724683854e-06,
"loss": 0.32520896196365356,
"step": 1153
},
{
"epoch": 1.580821917808219,
"grad_norm": 0.16675332188606262,
"learning_rate": 2.2313336197915292e-06,
"loss": 0.33812659978866577,
"step": 1154
},
{
"epoch": 1.5821917808219177,
"grad_norm": 0.15388154983520508,
"learning_rate": 2.2173827550443417e-06,
"loss": 0.3268328905105591,
"step": 1155
},
{
"epoch": 1.5835616438356164,
"grad_norm": 0.1577136367559433,
"learning_rate": 2.20347019911342e-06,
"loss": 0.32836878299713135,
"step": 1156
},
{
"epoch": 1.584931506849315,
"grad_norm": 0.1715579777956009,
"learning_rate": 2.1895960204813194e-06,
"loss": 0.3792486786842346,
"step": 1157
},
{
"epoch": 1.5863013698630137,
"grad_norm": 0.15956799685955048,
"learning_rate": 2.1757602874416993e-06,
"loss": 0.37752485275268555,
"step": 1158
},
{
"epoch": 1.5876712328767124,
"grad_norm": 0.16624705493450165,
"learning_rate": 2.1619630680989644e-06,
"loss": 0.3493725657463074,
"step": 1159
},
{
"epoch": 1.589041095890411,
"grad_norm": 0.16341926157474518,
"learning_rate": 2.148204430367954e-06,
"loss": 0.3541504144668579,
"step": 1160
},
{
"epoch": 1.5904109589041096,
"grad_norm": 0.15657103061676025,
"learning_rate": 2.1344844419735757e-06,
"loss": 0.3333907723426819,
"step": 1161
},
{
"epoch": 1.5917808219178082,
"grad_norm": 0.16000789403915405,
"learning_rate": 2.1208031704505118e-06,
"loss": 0.35130012035369873,
"step": 1162
},
{
"epoch": 1.5931506849315067,
"grad_norm": 0.1461535096168518,
"learning_rate": 2.107160683142856e-06,
"loss": 0.32430267333984375,
"step": 1163
},
{
"epoch": 1.5945205479452054,
"grad_norm": 0.14178651571273804,
"learning_rate": 2.0935570472037892e-06,
"loss": 0.3193168044090271,
"step": 1164
},
{
"epoch": 1.595890410958904,
"grad_norm": 0.14919495582580566,
"learning_rate": 2.079992329595263e-06,
"loss": 0.3251698613166809,
"step": 1165
},
{
"epoch": 1.5972602739726027,
"grad_norm": 0.16162943840026855,
"learning_rate": 2.0664665970876496e-06,
"loss": 0.3408505320549011,
"step": 1166
},
{
"epoch": 1.5986301369863014,
"grad_norm": 0.16736219823360443,
"learning_rate": 2.0529799162594242e-06,
"loss": 0.3512228727340698,
"step": 1167
},
{
"epoch": 1.6,
"grad_norm": 0.15167738497257233,
"learning_rate": 2.0395323534968412e-06,
"loss": 0.3489873707294464,
"step": 1168
},
{
"epoch": 1.6013698630136988,
"grad_norm": 0.15653009712696075,
"learning_rate": 2.0261239749935966e-06,
"loss": 0.31903886795043945,
"step": 1169
},
{
"epoch": 1.6027397260273972,
"grad_norm": 0.16038073599338531,
"learning_rate": 2.012754846750509e-06,
"loss": 0.35285457968711853,
"step": 1170
},
{
"epoch": 1.604109589041096,
"grad_norm": 0.1591605246067047,
"learning_rate": 1.9994250345751886e-06,
"loss": 0.32918378710746765,
"step": 1171
},
{
"epoch": 1.6054794520547944,
"grad_norm": 0.1398192048072815,
"learning_rate": 1.98613460408173e-06,
"loss": 0.31244635581970215,
"step": 1172
},
{
"epoch": 1.606849315068493,
"grad_norm": 0.155035600066185,
"learning_rate": 1.972883620690366e-06,
"loss": 0.36001700162887573,
"step": 1173
},
{
"epoch": 1.6082191780821917,
"grad_norm": 0.15247505903244019,
"learning_rate": 1.9596721496271565e-06,
"loss": 0.3517085611820221,
"step": 1174
},
{
"epoch": 1.6095890410958904,
"grad_norm": 0.15745189785957336,
"learning_rate": 1.9465002559236777e-06,
"loss": 0.3493137061595917,
"step": 1175
},
{
"epoch": 1.610958904109589,
"grad_norm": 0.15074560046195984,
"learning_rate": 1.9333680044166847e-06,
"loss": 0.33854442834854126,
"step": 1176
},
{
"epoch": 1.6123287671232878,
"grad_norm": 0.1640157252550125,
"learning_rate": 1.920275459747796e-06,
"loss": 0.3208191692829132,
"step": 1177
},
{
"epoch": 1.6136986301369864,
"grad_norm": 0.15582209825515747,
"learning_rate": 1.907222686363188e-06,
"loss": 0.3396856188774109,
"step": 1178
},
{
"epoch": 1.615068493150685,
"grad_norm": 0.1684257537126541,
"learning_rate": 1.8942097485132626e-06,
"loss": 0.34950849413871765,
"step": 1179
},
{
"epoch": 1.6164383561643836,
"grad_norm": 0.15507318079471588,
"learning_rate": 1.8812367102523365e-06,
"loss": 0.3066273331642151,
"step": 1180
},
{
"epoch": 1.617808219178082,
"grad_norm": 0.1489347517490387,
"learning_rate": 1.868303635438332e-06,
"loss": 0.33643442392349243,
"step": 1181
},
{
"epoch": 1.6191780821917807,
"grad_norm": 0.15879075229167938,
"learning_rate": 1.8554105877324525e-06,
"loss": 0.3287425637245178,
"step": 1182
},
{
"epoch": 1.6205479452054794,
"grad_norm": 0.17736220359802246,
"learning_rate": 1.8425576305988735e-06,
"loss": 0.37067461013793945,
"step": 1183
},
{
"epoch": 1.621917808219178,
"grad_norm": 0.1560935527086258,
"learning_rate": 1.8297448273044295e-06,
"loss": 0.35819360613822937,
"step": 1184
},
{
"epoch": 1.6232876712328768,
"grad_norm": 0.15256348252296448,
"learning_rate": 1.81697224091831e-06,
"loss": 0.3505377769470215,
"step": 1185
},
{
"epoch": 1.6246575342465754,
"grad_norm": 0.16711962223052979,
"learning_rate": 1.8042399343117368e-06,
"loss": 0.34088966250419617,
"step": 1186
},
{
"epoch": 1.6260273972602741,
"grad_norm": 0.14235368371009827,
"learning_rate": 1.7915479701576577e-06,
"loss": 0.30414897203445435,
"step": 1187
},
{
"epoch": 1.6273972602739726,
"grad_norm": 0.15138524770736694,
"learning_rate": 1.7788964109304495e-06,
"loss": 0.33105698227882385,
"step": 1188
},
{
"epoch": 1.6287671232876713,
"grad_norm": 0.1526041179895401,
"learning_rate": 1.7662853189055951e-06,
"loss": 0.33949634432792664,
"step": 1189
},
{
"epoch": 1.6301369863013697,
"grad_norm": 0.15941354632377625,
"learning_rate": 1.7537147561593826e-06,
"loss": 0.33900412917137146,
"step": 1190
},
{
"epoch": 1.6315068493150684,
"grad_norm": 0.15249750018119812,
"learning_rate": 1.7411847845686082e-06,
"loss": 0.32778996229171753,
"step": 1191
},
{
"epoch": 1.632876712328767,
"grad_norm": 0.15217413008213043,
"learning_rate": 1.7286954658102495e-06,
"loss": 0.3232656717300415,
"step": 1192
},
{
"epoch": 1.6342465753424658,
"grad_norm": 0.1474272459745407,
"learning_rate": 1.7162468613611937e-06,
"loss": 0.3270835280418396,
"step": 1193
},
{
"epoch": 1.6356164383561644,
"grad_norm": 0.16704101860523224,
"learning_rate": 1.703839032497906e-06,
"loss": 0.3511890470981598,
"step": 1194
},
{
"epoch": 1.6369863013698631,
"grad_norm": 0.1573154777288437,
"learning_rate": 1.6914720402961403e-06,
"loss": 0.35664379596710205,
"step": 1195
},
{
"epoch": 1.6383561643835618,
"grad_norm": 0.1522405743598938,
"learning_rate": 1.6791459456306457e-06,
"loss": 0.31690123677253723,
"step": 1196
},
{
"epoch": 1.6397260273972603,
"grad_norm": 0.14940029382705688,
"learning_rate": 1.6668608091748495e-06,
"loss": 0.32694342732429504,
"step": 1197
},
{
"epoch": 1.641095890410959,
"grad_norm": 0.16398581862449646,
"learning_rate": 1.6546166914005768e-06,
"loss": 0.3361591100692749,
"step": 1198
},
{
"epoch": 1.6424657534246574,
"grad_norm": 0.15232853591442108,
"learning_rate": 1.6424136525777367e-06,
"loss": 0.3171868324279785,
"step": 1199
},
{
"epoch": 1.643835616438356,
"grad_norm": 0.16350358724594116,
"learning_rate": 1.6302517527740358e-06,
"loss": 0.35954880714416504,
"step": 1200
},
{
"epoch": 1.6452054794520548,
"grad_norm": 0.16470475494861603,
"learning_rate": 1.6181310518546856e-06,
"loss": 0.342126727104187,
"step": 1201
},
{
"epoch": 1.6465753424657534,
"grad_norm": 0.15912854671478271,
"learning_rate": 1.6060516094820876e-06,
"loss": 0.32483920454978943,
"step": 1202
},
{
"epoch": 1.6479452054794521,
"grad_norm": 0.16000115871429443,
"learning_rate": 1.5940134851155698e-06,
"loss": 0.33072608709335327,
"step": 1203
},
{
"epoch": 1.6493150684931508,
"grad_norm": 0.17337526381015778,
"learning_rate": 1.5820167380110697e-06,
"loss": 0.3809818625450134,
"step": 1204
},
{
"epoch": 1.6506849315068495,
"grad_norm": 0.1659628003835678,
"learning_rate": 1.5700614272208492e-06,
"loss": 0.3534291982650757,
"step": 1205
},
{
"epoch": 1.652054794520548,
"grad_norm": 0.15824678540229797,
"learning_rate": 1.558147611593216e-06,
"loss": 0.34972262382507324,
"step": 1206
},
{
"epoch": 1.6534246575342466,
"grad_norm": 0.15221920609474182,
"learning_rate": 1.5462753497722139e-06,
"loss": 0.32763582468032837,
"step": 1207
},
{
"epoch": 1.654794520547945,
"grad_norm": 0.15916042029857635,
"learning_rate": 1.5344447001973462e-06,
"loss": 0.31656497716903687,
"step": 1208
},
{
"epoch": 1.6561643835616437,
"grad_norm": 0.1665540486574173,
"learning_rate": 1.522655721103291e-06,
"loss": 0.37915846705436707,
"step": 1209
},
{
"epoch": 1.6575342465753424,
"grad_norm": 0.1595245897769928,
"learning_rate": 1.5109084705196041e-06,
"loss": 0.3317162096500397,
"step": 1210
},
{
"epoch": 1.658904109589041,
"grad_norm": 0.1856878399848938,
"learning_rate": 1.49920300627044e-06,
"loss": 0.33943232893943787,
"step": 1211
},
{
"epoch": 1.6602739726027398,
"grad_norm": 0.15331265330314636,
"learning_rate": 1.4875393859742626e-06,
"loss": 0.34050869941711426,
"step": 1212
},
{
"epoch": 1.6616438356164385,
"grad_norm": 0.1600010246038437,
"learning_rate": 1.475917667043575e-06,
"loss": 0.3265448212623596,
"step": 1213
},
{
"epoch": 1.6630136986301371,
"grad_norm": 0.16359809041023254,
"learning_rate": 1.4643379066846186e-06,
"loss": 0.33075857162475586,
"step": 1214
},
{
"epoch": 1.6643835616438356,
"grad_norm": 0.1521827131509781,
"learning_rate": 1.4528001618970966e-06,
"loss": 0.31874728202819824,
"step": 1215
},
{
"epoch": 1.6657534246575343,
"grad_norm": 0.1428581029176712,
"learning_rate": 1.4413044894739092e-06,
"loss": 0.3019358515739441,
"step": 1216
},
{
"epoch": 1.6671232876712327,
"grad_norm": 0.16607871651649475,
"learning_rate": 1.4298509460008491e-06,
"loss": 0.31903016567230225,
"step": 1217
},
{
"epoch": 1.6684931506849314,
"grad_norm": 0.15393413603305817,
"learning_rate": 1.4184395878563395e-06,
"loss": 0.33354055881500244,
"step": 1218
},
{
"epoch": 1.66986301369863,
"grad_norm": 0.15508490800857544,
"learning_rate": 1.4070704712111572e-06,
"loss": 0.32358676195144653,
"step": 1219
},
{
"epoch": 1.6712328767123288,
"grad_norm": 0.14650438725948334,
"learning_rate": 1.395743652028143e-06,
"loss": 0.33141303062438965,
"step": 1220
},
{
"epoch": 1.6726027397260275,
"grad_norm": 0.15726953744888306,
"learning_rate": 1.3844591860619382e-06,
"loss": 0.35107171535491943,
"step": 1221
},
{
"epoch": 1.6739726027397261,
"grad_norm": 0.15117298066616058,
"learning_rate": 1.3732171288587105e-06,
"loss": 0.30023708939552307,
"step": 1222
},
{
"epoch": 1.6753424657534246,
"grad_norm": 0.1782534420490265,
"learning_rate": 1.3620175357558706e-06,
"loss": 0.34968823194503784,
"step": 1223
},
{
"epoch": 1.6767123287671233,
"grad_norm": 0.17389066517353058,
"learning_rate": 1.3508604618818067e-06,
"loss": 0.31941914558410645,
"step": 1224
},
{
"epoch": 1.678082191780822,
"grad_norm": 0.17911015450954437,
"learning_rate": 1.339745962155613e-06,
"loss": 0.3240607678890228,
"step": 1225
},
{
"epoch": 1.6794520547945204,
"grad_norm": 0.1515609174966812,
"learning_rate": 1.3286740912868234e-06,
"loss": 0.3442304730415344,
"step": 1226
},
{
"epoch": 1.680821917808219,
"grad_norm": 0.15644872188568115,
"learning_rate": 1.3176449037751294e-06,
"loss": 0.31893715262413025,
"step": 1227
},
{
"epoch": 1.6821917808219178,
"grad_norm": 0.1573629379272461,
"learning_rate": 1.3066584539101224e-06,
"loss": 0.3529011011123657,
"step": 1228
},
{
"epoch": 1.6835616438356165,
"grad_norm": 0.17355655133724213,
"learning_rate": 1.2957147957710292e-06,
"loss": 0.3373495638370514,
"step": 1229
},
{
"epoch": 1.6849315068493151,
"grad_norm": 0.151940256357193,
"learning_rate": 1.2848139832264328e-06,
"loss": 0.34251755475997925,
"step": 1230
},
{
"epoch": 1.6863013698630138,
"grad_norm": 0.15687845647335052,
"learning_rate": 1.273956069934017e-06,
"loss": 0.33196067810058594,
"step": 1231
},
{
"epoch": 1.6876712328767123,
"grad_norm": 0.16004379093647003,
"learning_rate": 1.263141109340308e-06,
"loss": 0.3280409872531891,
"step": 1232
},
{
"epoch": 1.689041095890411,
"grad_norm": 0.1561116725206375,
"learning_rate": 1.2523691546803872e-06,
"loss": 0.36071187257766724,
"step": 1233
},
{
"epoch": 1.6904109589041096,
"grad_norm": 0.16006499528884888,
"learning_rate": 1.2416402589776632e-06,
"loss": 0.3473643660545349,
"step": 1234
},
{
"epoch": 1.691780821917808,
"grad_norm": 0.1759638637304306,
"learning_rate": 1.230954475043581e-06,
"loss": 0.367775022983551,
"step": 1235
},
{
"epoch": 1.6931506849315068,
"grad_norm": 0.1692914068698883,
"learning_rate": 1.2203118554773807e-06,
"loss": 0.3360450267791748,
"step": 1236
},
{
"epoch": 1.6945205479452055,
"grad_norm": 0.15737596154212952,
"learning_rate": 1.2097124526658277e-06,
"loss": 0.3283913731575012,
"step": 1237
},
{
"epoch": 1.6958904109589041,
"grad_norm": 0.19927242398262024,
"learning_rate": 1.1991563187829592e-06,
"loss": 0.34028875827789307,
"step": 1238
},
{
"epoch": 1.6972602739726028,
"grad_norm": 0.15856406092643738,
"learning_rate": 1.1886435057898338e-06,
"loss": 0.34818997979164124,
"step": 1239
},
{
"epoch": 1.6986301369863015,
"grad_norm": 0.1475028097629547,
"learning_rate": 1.1781740654342622e-06,
"loss": 0.3465500473976135,
"step": 1240
},
{
"epoch": 1.7,
"grad_norm": 0.16196544468402863,
"learning_rate": 1.1677480492505589e-06,
"loss": 0.3488352298736572,
"step": 1241
},
{
"epoch": 1.7013698630136986,
"grad_norm": 0.14074058830738068,
"learning_rate": 1.1573655085592983e-06,
"loss": 0.29739153385162354,
"step": 1242
},
{
"epoch": 1.7027397260273973,
"grad_norm": 0.16181747615337372,
"learning_rate": 1.1470264944670385e-06,
"loss": 0.3437436521053314,
"step": 1243
},
{
"epoch": 1.7041095890410958,
"grad_norm": 0.15629208087921143,
"learning_rate": 1.1367310578660962e-06,
"loss": 0.3288268446922302,
"step": 1244
},
{
"epoch": 1.7054794520547945,
"grad_norm": 0.15012064576148987,
"learning_rate": 1.1264792494342858e-06,
"loss": 0.333500474691391,
"step": 1245
},
{
"epoch": 1.7068493150684931,
"grad_norm": 0.17188264429569244,
"learning_rate": 1.1162711196346532e-06,
"loss": 0.38367658853530884,
"step": 1246
},
{
"epoch": 1.7082191780821918,
"grad_norm": 0.1565122753381729,
"learning_rate": 1.1061067187152584e-06,
"loss": 0.31294959783554077,
"step": 1247
},
{
"epoch": 1.7095890410958905,
"grad_norm": 0.1444123089313507,
"learning_rate": 1.0959860967089042e-06,
"loss": 0.29740363359451294,
"step": 1248
},
{
"epoch": 1.7109589041095892,
"grad_norm": 0.16455943882465363,
"learning_rate": 1.0859093034328972e-06,
"loss": 0.3299975097179413,
"step": 1249
},
{
"epoch": 1.7123287671232876,
"grad_norm": 0.14738664031028748,
"learning_rate": 1.07587638848881e-06,
"loss": 0.3134653866291046,
"step": 1250
},
{
"epoch": 1.7136986301369863,
"grad_norm": 0.15507303178310394,
"learning_rate": 1.0658874012622244e-06,
"loss": 0.345506489276886,
"step": 1251
},
{
"epoch": 1.7150684931506848,
"grad_norm": 0.1389225721359253,
"learning_rate": 1.0559423909224942e-06,
"loss": 0.2806209921836853,
"step": 1252
},
{
"epoch": 1.7164383561643834,
"grad_norm": 0.14970001578330994,
"learning_rate": 1.0460414064225099e-06,
"loss": 0.3479157090187073,
"step": 1253
},
{
"epoch": 1.7178082191780821,
"grad_norm": 0.17388109862804413,
"learning_rate": 1.0361844964984435e-06,
"loss": 0.36045724153518677,
"step": 1254
},
{
"epoch": 1.7191780821917808,
"grad_norm": 0.14937163889408112,
"learning_rate": 1.0263717096695192e-06,
"loss": 0.33035242557525635,
"step": 1255
},
{
"epoch": 1.7205479452054795,
"grad_norm": 0.16189490258693695,
"learning_rate": 1.0166030942377703e-06,
"loss": 0.35016894340515137,
"step": 1256
},
{
"epoch": 1.7219178082191782,
"grad_norm": 0.15496496856212616,
"learning_rate": 1.0068786982878087e-06,
"loss": 0.35125821828842163,
"step": 1257
},
{
"epoch": 1.7232876712328768,
"grad_norm": 0.15979105234146118,
"learning_rate": 9.971985696865737e-07,
"loss": 0.34419381618499756,
"step": 1258
},
{
"epoch": 1.7246575342465753,
"grad_norm": 0.17270156741142273,
"learning_rate": 9.87562756083108e-07,
"loss": 0.38739997148513794,
"step": 1259
},
{
"epoch": 1.726027397260274,
"grad_norm": 0.16050505638122559,
"learning_rate": 9.77971304908325e-07,
"loss": 0.36073169112205505,
"step": 1260
},
{
"epoch": 1.7273972602739724,
"grad_norm": 0.15811535716056824,
"learning_rate": 9.684242633747642e-07,
"loss": 0.3458669185638428,
"step": 1261
},
{
"epoch": 1.7287671232876711,
"grad_norm": 0.16361944377422333,
"learning_rate": 9.589216784763655e-07,
"loss": 0.35864222049713135,
"step": 1262
},
{
"epoch": 1.7301369863013698,
"grad_norm": 0.16463346779346466,
"learning_rate": 9.494635969882426e-07,
"loss": 0.3516255021095276,
"step": 1263
},
{
"epoch": 1.7315068493150685,
"grad_norm": 0.13944090902805328,
"learning_rate": 9.400500654664413e-07,
"loss": 0.3028295338153839,
"step": 1264
},
{
"epoch": 1.7328767123287672,
"grad_norm": 0.15407785773277283,
"learning_rate": 9.306811302477214e-07,
"loss": 0.3290501534938812,
"step": 1265
},
{
"epoch": 1.7342465753424658,
"grad_norm": 0.17160239815711975,
"learning_rate": 9.213568374493176e-07,
"loss": 0.3213893175125122,
"step": 1266
},
{
"epoch": 1.7356164383561645,
"grad_norm": 0.1475364714860916,
"learning_rate": 9.120772329687278e-07,
"loss": 0.3017924427986145,
"step": 1267
},
{
"epoch": 1.736986301369863,
"grad_norm": 0.1735687404870987,
"learning_rate": 9.028423624834703e-07,
"loss": 0.34012332558631897,
"step": 1268
},
{
"epoch": 1.7383561643835617,
"grad_norm": 0.14973114430904388,
"learning_rate": 8.936522714508678e-07,
"loss": 0.3517354130744934,
"step": 1269
},
{
"epoch": 1.7397260273972601,
"grad_norm": 0.15395504236221313,
"learning_rate": 8.845070051078286e-07,
"loss": 0.31232795119285583,
"step": 1270
},
{
"epoch": 1.7410958904109588,
"grad_norm": 0.1512281894683838,
"learning_rate": 8.754066084706125e-07,
"loss": 0.3256188929080963,
"step": 1271
},
{
"epoch": 1.7424657534246575,
"grad_norm": 0.1710251122713089,
"learning_rate": 8.663511263346114e-07,
"loss": 0.3719729781150818,
"step": 1272
},
{
"epoch": 1.7438356164383562,
"grad_norm": 0.1589270681142807,
"learning_rate": 8.573406032741438e-07,
"loss": 0.32557862997055054,
"step": 1273
},
{
"epoch": 1.7452054794520548,
"grad_norm": 0.16728456318378448,
"learning_rate": 8.483750836422078e-07,
"loss": 0.30554038286209106,
"step": 1274
},
{
"epoch": 1.7465753424657535,
"grad_norm": 0.15205995738506317,
"learning_rate": 8.394546115702928e-07,
"loss": 0.3218629062175751,
"step": 1275
},
{
"epoch": 1.7479452054794522,
"grad_norm": 0.24099785089492798,
"learning_rate": 8.305792309681382e-07,
"loss": 0.3274707794189453,
"step": 1276
},
{
"epoch": 1.7493150684931507,
"grad_norm": 0.17904773354530334,
"learning_rate": 8.217489855235338e-07,
"loss": 0.35155755281448364,
"step": 1277
},
{
"epoch": 1.7506849315068493,
"grad_norm": 0.17398542165756226,
"learning_rate": 8.129639187020954e-07,
"loss": 0.3349018692970276,
"step": 1278
},
{
"epoch": 1.7520547945205478,
"grad_norm": 0.15470083057880402,
"learning_rate": 8.042240737470508e-07,
"loss": 0.30421948432922363,
"step": 1279
},
{
"epoch": 1.7534246575342465,
"grad_norm": 0.14305900037288666,
"learning_rate": 7.955294936790359e-07,
"loss": 0.31143707036972046,
"step": 1280
},
{
"epoch": 1.7547945205479452,
"grad_norm": 0.14751826226711273,
"learning_rate": 7.868802212958704e-07,
"loss": 0.2954302132129669,
"step": 1281
},
{
"epoch": 1.7561643835616438,
"grad_norm": 0.1575712412595749,
"learning_rate": 7.782762991723558e-07,
"loss": 0.30905646085739136,
"step": 1282
},
{
"epoch": 1.7575342465753425,
"grad_norm": 0.14383475482463837,
"learning_rate": 7.697177696600677e-07,
"loss": 0.32206839323043823,
"step": 1283
},
{
"epoch": 1.7589041095890412,
"grad_norm": 0.15969853103160858,
"learning_rate": 7.612046748871327e-07,
"loss": 0.32025647163391113,
"step": 1284
},
{
"epoch": 1.7602739726027399,
"grad_norm": 0.15641602873802185,
"learning_rate": 7.527370567580416e-07,
"loss": 0.3522297143936157,
"step": 1285
},
{
"epoch": 1.7616438356164383,
"grad_norm": 0.14964739978313446,
"learning_rate": 7.443149569534325e-07,
"loss": 0.3237984776496887,
"step": 1286
},
{
"epoch": 1.763013698630137,
"grad_norm": 0.1524275839328766,
"learning_rate": 7.359384169298744e-07,
"loss": 0.31260764598846436,
"step": 1287
},
{
"epoch": 1.7643835616438355,
"grad_norm": 0.16798636317253113,
"learning_rate": 7.276074779196906e-07,
"loss": 0.37301307916641235,
"step": 1288
},
{
"epoch": 1.7657534246575342,
"grad_norm": 0.15152223408222198,
"learning_rate": 7.193221809307304e-07,
"loss": 0.30456385016441345,
"step": 1289
},
{
"epoch": 1.7671232876712328,
"grad_norm": 0.16104163229465485,
"learning_rate": 7.110825667461762e-07,
"loss": 0.35112541913986206,
"step": 1290
},
{
"epoch": 1.7684931506849315,
"grad_norm": 0.1392669379711151,
"learning_rate": 7.028886759243503e-07,
"loss": 0.2964078485965729,
"step": 1291
},
{
"epoch": 1.7698630136986302,
"grad_norm": 0.15071091055870056,
"learning_rate": 6.947405487985015e-07,
"loss": 0.28317034244537354,
"step": 1292
},
{
"epoch": 1.7712328767123289,
"grad_norm": 0.1536220759153366,
"learning_rate": 6.866382254766158e-07,
"loss": 0.3215874433517456,
"step": 1293
},
{
"epoch": 1.7726027397260276,
"grad_norm": 0.1379450261592865,
"learning_rate": 6.785817458412181e-07,
"loss": 0.2879907190799713,
"step": 1294
},
{
"epoch": 1.773972602739726,
"grad_norm": 0.17187511920928955,
"learning_rate": 6.705711495491718e-07,
"loss": 0.3147532343864441,
"step": 1295
},
{
"epoch": 1.7753424657534247,
"grad_norm": 0.17327673733234406,
"learning_rate": 6.62606476031491e-07,
"loss": 0.3378913104534149,
"step": 1296
},
{
"epoch": 1.7767123287671232,
"grad_norm": 0.15394513309001923,
"learning_rate": 6.546877644931315e-07,
"loss": 0.32351988554000854,
"step": 1297
},
{
"epoch": 1.7780821917808218,
"grad_norm": 0.14969897270202637,
"learning_rate": 6.468150539128193e-07,
"loss": 0.3032512068748474,
"step": 1298
},
{
"epoch": 1.7794520547945205,
"grad_norm": 0.15381179749965668,
"learning_rate": 6.3898838304284e-07,
"loss": 0.3199779689311981,
"step": 1299
},
{
"epoch": 1.7808219178082192,
"grad_norm": 0.15317706763744354,
"learning_rate": 6.312077904088565e-07,
"loss": 0.32301846146583557,
"step": 1300
},
{
"epoch": 1.7821917808219179,
"grad_norm": 0.152884840965271,
"learning_rate": 6.234733143097215e-07,
"loss": 0.29910945892333984,
"step": 1301
},
{
"epoch": 1.7835616438356166,
"grad_norm": 0.1553654968738556,
"learning_rate": 6.157849928172832e-07,
"loss": 0.3187806010246277,
"step": 1302
},
{
"epoch": 1.7849315068493152,
"grad_norm": 0.16176237165927887,
"learning_rate": 6.081428637761977e-07,
"loss": 0.339767187833786,
"step": 1303
},
{
"epoch": 1.7863013698630137,
"grad_norm": 0.15177124738693237,
"learning_rate": 6.005469648037532e-07,
"loss": 0.3246235251426697,
"step": 1304
},
{
"epoch": 1.7876712328767124,
"grad_norm": 0.1596703827381134,
"learning_rate": 5.929973332896677e-07,
"loss": 0.35482412576675415,
"step": 1305
},
{
"epoch": 1.7890410958904108,
"grad_norm": 0.14820478856563568,
"learning_rate": 5.854940063959213e-07,
"loss": 0.3279823064804077,
"step": 1306
},
{
"epoch": 1.7904109589041095,
"grad_norm": 0.14746461808681488,
"learning_rate": 5.7803702105656e-07,
"loss": 0.3030048608779907,
"step": 1307
},
{
"epoch": 1.7917808219178082,
"grad_norm": 0.15053853392601013,
"learning_rate": 5.706264139775286e-07,
"loss": 0.31790563464164734,
"step": 1308
},
{
"epoch": 1.7931506849315069,
"grad_norm": 0.15882104635238647,
"learning_rate": 5.632622216364736e-07,
"loss": 0.32655006647109985,
"step": 1309
},
{
"epoch": 1.7945205479452055,
"grad_norm": 0.16202262043952942,
"learning_rate": 5.559444802825741e-07,
"loss": 0.3297027349472046,
"step": 1310
},
{
"epoch": 1.7958904109589042,
"grad_norm": 0.16546350717544556,
"learning_rate": 5.486732259363647e-07,
"loss": 0.3207970857620239,
"step": 1311
},
{
"epoch": 1.7972602739726027,
"grad_norm": 0.16333724558353424,
"learning_rate": 5.414484943895492e-07,
"loss": 0.33857011795043945,
"step": 1312
},
{
"epoch": 1.7986301369863014,
"grad_norm": 0.1690317839384079,
"learning_rate": 5.342703212048306e-07,
"loss": 0.38763177394866943,
"step": 1313
},
{
"epoch": 1.8,
"grad_norm": 0.14536961913108826,
"learning_rate": 5.271387417157392e-07,
"loss": 0.31603944301605225,
"step": 1314
},
{
"epoch": 1.8013698630136985,
"grad_norm": 0.16719646751880646,
"learning_rate": 5.200537910264481e-07,
"loss": 0.35231393575668335,
"step": 1315
},
{
"epoch": 1.8027397260273972,
"grad_norm": 0.17478609085083008,
"learning_rate": 5.130155040116125e-07,
"loss": 0.35779231786727905,
"step": 1316
},
{
"epoch": 1.8041095890410959,
"grad_norm": 0.15386946499347687,
"learning_rate": 5.060239153161872e-07,
"loss": 0.32011014223098755,
"step": 1317
},
{
"epoch": 1.8054794520547945,
"grad_norm": 0.1588115394115448,
"learning_rate": 4.990790593552663e-07,
"loss": 0.294185996055603,
"step": 1318
},
{
"epoch": 1.8068493150684932,
"grad_norm": 0.16609877347946167,
"learning_rate": 4.921809703139068e-07,
"loss": 0.30233830213546753,
"step": 1319
},
{
"epoch": 1.808219178082192,
"grad_norm": 0.1439180076122284,
"learning_rate": 4.853296821469589e-07,
"loss": 0.2957511842250824,
"step": 1320
},
{
"epoch": 1.8095890410958904,
"grad_norm": 0.16486167907714844,
"learning_rate": 4.785252285789077e-07,
"loss": 0.3310033082962036,
"step": 1321
},
{
"epoch": 1.810958904109589,
"grad_norm": 0.17464743554592133,
"learning_rate": 4.7176764310369994e-07,
"loss": 0.3550257086753845,
"step": 1322
},
{
"epoch": 1.8123287671232877,
"grad_norm": 0.17432478070259094,
"learning_rate": 4.650569589845766e-07,
"loss": 0.34577667713165283,
"step": 1323
},
{
"epoch": 1.8136986301369862,
"grad_norm": 0.14168258011341095,
"learning_rate": 4.5839320925392117e-07,
"loss": 0.30111217498779297,
"step": 1324
},
{
"epoch": 1.8150684931506849,
"grad_norm": 0.16613982617855072,
"learning_rate": 4.5177642671308087e-07,
"loss": 0.3328065276145935,
"step": 1325
},
{
"epoch": 1.8164383561643835,
"grad_norm": 0.16824592649936676,
"learning_rate": 4.4520664393222e-07,
"loss": 0.3466789126396179,
"step": 1326
},
{
"epoch": 1.8178082191780822,
"grad_norm": 0.15729574859142303,
"learning_rate": 4.386838932501547e-07,
"loss": 0.32296979427337646,
"step": 1327
},
{
"epoch": 1.819178082191781,
"grad_norm": 0.17344285547733307,
"learning_rate": 4.3220820677418285e-07,
"loss": 0.3270629048347473,
"step": 1328
},
{
"epoch": 1.8205479452054796,
"grad_norm": 0.17102137207984924,
"learning_rate": 4.2577961637994544e-07,
"loss": 0.35478442907333374,
"step": 1329
},
{
"epoch": 1.821917808219178,
"grad_norm": 0.1614276021718979,
"learning_rate": 4.193981537112557e-07,
"loss": 0.35880762338638306,
"step": 1330
},
{
"epoch": 1.8232876712328767,
"grad_norm": 0.15321889519691467,
"learning_rate": 4.1306385017994443e-07,
"loss": 0.325344443321228,
"step": 1331
},
{
"epoch": 1.8246575342465754,
"grad_norm": 0.1599494367837906,
"learning_rate": 4.067767369657161e-07,
"loss": 0.3106187880039215,
"step": 1332
},
{
"epoch": 1.8260273972602739,
"grad_norm": 0.15317757427692413,
"learning_rate": 4.0053684501598076e-07,
"loss": 0.3269118070602417,
"step": 1333
},
{
"epoch": 1.8273972602739725,
"grad_norm": 0.16627809405326843,
"learning_rate": 3.9434420504571027e-07,
"loss": 0.35508206486701965,
"step": 1334
},
{
"epoch": 1.8287671232876712,
"grad_norm": 0.153465136885643,
"learning_rate": 3.8819884753728665e-07,
"loss": 0.30518004298210144,
"step": 1335
},
{
"epoch": 1.83013698630137,
"grad_norm": 0.1573999524116516,
"learning_rate": 3.821008027403494e-07,
"loss": 0.3222675919532776,
"step": 1336
},
{
"epoch": 1.8315068493150686,
"grad_norm": 0.1496000587940216,
"learning_rate": 3.7605010067165216e-07,
"loss": 0.3345215320587158,
"step": 1337
},
{
"epoch": 1.8328767123287673,
"grad_norm": 0.15938860177993774,
"learning_rate": 3.700467711149025e-07,
"loss": 0.3547212481498718,
"step": 1338
},
{
"epoch": 1.8342465753424657,
"grad_norm": 0.14790597558021545,
"learning_rate": 3.6409084362063143e-07,
"loss": 0.3036991059780121,
"step": 1339
},
{
"epoch": 1.8356164383561644,
"grad_norm": 0.16490647196769714,
"learning_rate": 3.581823475060364e-07,
"loss": 0.31740397214889526,
"step": 1340
},
{
"epoch": 1.8369863013698629,
"grad_norm": 0.16203896701335907,
"learning_rate": 3.5232131185484075e-07,
"loss": 0.35205888748168945,
"step": 1341
},
{
"epoch": 1.8383561643835615,
"grad_norm": 0.16094939410686493,
"learning_rate": 3.4650776551715225e-07,
"loss": 0.3516027331352234,
"step": 1342
},
{
"epoch": 1.8397260273972602,
"grad_norm": 0.1690695583820343,
"learning_rate": 3.4074173710931804e-07,
"loss": 0.3475687801837921,
"step": 1343
},
{
"epoch": 1.841095890410959,
"grad_norm": 0.1640596091747284,
"learning_rate": 3.350232550137833e-07,
"loss": 0.33157050609588623,
"step": 1344
},
{
"epoch": 1.8424657534246576,
"grad_norm": 0.1597766876220703,
"learning_rate": 3.2935234737895837e-07,
"loss": 0.3033367395401001,
"step": 1345
},
{
"epoch": 1.8438356164383563,
"grad_norm": 0.1684199720621109,
"learning_rate": 3.23729042119072e-07,
"loss": 0.34482163190841675,
"step": 1346
},
{
"epoch": 1.845205479452055,
"grad_norm": 0.1747860610485077,
"learning_rate": 3.1815336691403464e-07,
"loss": 0.3802839517593384,
"step": 1347
},
{
"epoch": 1.8465753424657534,
"grad_norm": 0.15639357268810272,
"learning_rate": 3.1262534920930785e-07,
"loss": 0.3377276062965393,
"step": 1348
},
{
"epoch": 1.847945205479452,
"grad_norm": 0.15243448317050934,
"learning_rate": 3.071450162157663e-07,
"loss": 0.31236010789871216,
"step": 1349
},
{
"epoch": 1.8493150684931505,
"grad_norm": 0.1585215926170349,
"learning_rate": 3.0171239490956237e-07,
"loss": 0.3301020562648773,
"step": 1350
},
{
"epoch": 1.8506849315068492,
"grad_norm": 0.1702500581741333,
"learning_rate": 2.9632751203199305e-07,
"loss": 0.34906840324401855,
"step": 1351
},
{
"epoch": 1.8520547945205479,
"grad_norm": 0.1580081433057785,
"learning_rate": 2.9099039408937435e-07,
"loss": 0.31397759914398193,
"step": 1352
},
{
"epoch": 1.8534246575342466,
"grad_norm": 0.177195206284523,
"learning_rate": 2.857010673529015e-07,
"loss": 0.35444584488868713,
"step": 1353
},
{
"epoch": 1.8547945205479452,
"grad_norm": 0.15439055860042572,
"learning_rate": 2.8045955785852453e-07,
"loss": 0.33671659231185913,
"step": 1354
},
{
"epoch": 1.856164383561644,
"grad_norm": 0.17044661939144135,
"learning_rate": 2.75265891406824e-07,
"loss": 0.3560411334037781,
"step": 1355
},
{
"epoch": 1.8575342465753426,
"grad_norm": 0.16385668516159058,
"learning_rate": 2.701200935628767e-07,
"loss": 0.315941721200943,
"step": 1356
},
{
"epoch": 1.858904109589041,
"grad_norm": 0.15318404138088226,
"learning_rate": 2.6502218965613335e-07,
"loss": 0.3170655369758606,
"step": 1357
},
{
"epoch": 1.8602739726027397,
"grad_norm": 0.1832403540611267,
"learning_rate": 2.599722047802955e-07,
"loss": 0.37524086236953735,
"step": 1358
},
{
"epoch": 1.8616438356164382,
"grad_norm": 0.1468266397714615,
"learning_rate": 2.5497016379318894e-07,
"loss": 0.3305490016937256,
"step": 1359
},
{
"epoch": 1.8630136986301369,
"grad_norm": 0.16064555943012238,
"learning_rate": 2.500160913166405e-07,
"loss": 0.34337884187698364,
"step": 1360
},
{
"epoch": 1.8643835616438356,
"grad_norm": 0.1646057367324829,
"learning_rate": 2.451100117363603e-07,
"loss": 0.3477829396724701,
"step": 1361
},
{
"epoch": 1.8657534246575342,
"grad_norm": 0.1530703604221344,
"learning_rate": 2.4025194920182405e-07,
"loss": 0.32335418462753296,
"step": 1362
},
{
"epoch": 1.867123287671233,
"grad_norm": 0.1550067663192749,
"learning_rate": 2.354419276261466e-07,
"loss": 0.3232346773147583,
"step": 1363
},
{
"epoch": 1.8684931506849316,
"grad_norm": 0.17549198865890503,
"learning_rate": 2.3067997068596526e-07,
"loss": 0.3539588749408722,
"step": 1364
},
{
"epoch": 1.8698630136986303,
"grad_norm": 0.1565999686717987,
"learning_rate": 2.2596610182133328e-07,
"loss": 0.36170855164527893,
"step": 1365
},
{
"epoch": 1.8712328767123287,
"grad_norm": 0.14914360642433167,
"learning_rate": 2.2130034423559322e-07,
"loss": 0.3174242377281189,
"step": 1366
},
{
"epoch": 1.8726027397260274,
"grad_norm": 0.15867829322814941,
"learning_rate": 2.1668272089526377e-07,
"loss": 0.3288201689720154,
"step": 1367
},
{
"epoch": 1.8739726027397259,
"grad_norm": 0.15549346804618835,
"learning_rate": 2.121132545299376e-07,
"loss": 0.3385423421859741,
"step": 1368
},
{
"epoch": 1.8753424657534246,
"grad_norm": 0.15481318533420563,
"learning_rate": 2.0759196763215362e-07,
"loss": 0.3591095805168152,
"step": 1369
},
{
"epoch": 1.8767123287671232,
"grad_norm": 0.16956226527690887,
"learning_rate": 2.0311888245729827e-07,
"loss": 0.3723497688770294,
"step": 1370
},
{
"epoch": 1.878082191780822,
"grad_norm": 0.17633791267871857,
"learning_rate": 1.986940210234922e-07,
"loss": 0.37217652797698975,
"step": 1371
},
{
"epoch": 1.8794520547945206,
"grad_norm": 0.1783612221479416,
"learning_rate": 1.9431740511147935e-07,
"loss": 0.36250248551368713,
"step": 1372
},
{
"epoch": 1.8808219178082193,
"grad_norm": 0.14923235774040222,
"learning_rate": 1.8998905626452568e-07,
"loss": 0.3103126287460327,
"step": 1373
},
{
"epoch": 1.882191780821918,
"grad_norm": 0.1442461758852005,
"learning_rate": 1.8570899578830293e-07,
"loss": 0.2998763918876648,
"step": 1374
},
{
"epoch": 1.8835616438356164,
"grad_norm": 0.15582527220249176,
"learning_rate": 1.8147724475079954e-07,
"loss": 0.32586419582366943,
"step": 1375
},
{
"epoch": 1.884931506849315,
"grad_norm": 0.15638771653175354,
"learning_rate": 1.7729382398219862e-07,
"loss": 0.29698312282562256,
"step": 1376
},
{
"epoch": 1.8863013698630136,
"grad_norm": 0.1632920205593109,
"learning_rate": 1.731587540747903e-07,
"loss": 0.37994974851608276,
"step": 1377
},
{
"epoch": 1.8876712328767122,
"grad_norm": 0.1591259092092514,
"learning_rate": 1.6907205538286287e-07,
"loss": 0.3357952833175659,
"step": 1378
},
{
"epoch": 1.889041095890411,
"grad_norm": 0.14531132578849792,
"learning_rate": 1.6503374802260274e-07,
"loss": 0.28875723481178284,
"step": 1379
},
{
"epoch": 1.8904109589041096,
"grad_norm": 0.1615014225244522,
"learning_rate": 1.6104385187199812e-07,
"loss": 0.32802480459213257,
"step": 1380
},
{
"epoch": 1.8917808219178083,
"grad_norm": 0.1593087762594223,
"learning_rate": 1.5710238657074218e-07,
"loss": 0.36830419301986694,
"step": 1381
},
{
"epoch": 1.893150684931507,
"grad_norm": 0.16016696393489838,
"learning_rate": 1.5320937152012881e-07,
"loss": 0.3358847200870514,
"step": 1382
},
{
"epoch": 1.8945205479452056,
"grad_norm": 0.15191650390625,
"learning_rate": 1.4936482588296942e-07,
"loss": 0.30469363927841187,
"step": 1383
},
{
"epoch": 1.895890410958904,
"grad_norm": 0.1591327041387558,
"learning_rate": 1.4556876858348611e-07,
"loss": 0.34513309597969055,
"step": 1384
},
{
"epoch": 1.8972602739726028,
"grad_norm": 0.15571986138820648,
"learning_rate": 1.418212183072254e-07,
"loss": 0.3193632960319519,
"step": 1385
},
{
"epoch": 1.8986301369863012,
"grad_norm": 0.1573391556739807,
"learning_rate": 1.381221935009669e-07,
"loss": 0.3403557240962982,
"step": 1386
},
{
"epoch": 1.9,
"grad_norm": 0.1758800446987152,
"learning_rate": 1.3447171237262912e-07,
"loss": 0.3336338400840759,
"step": 1387
},
{
"epoch": 1.9013698630136986,
"grad_norm": 0.15732204914093018,
"learning_rate": 1.3086979289118173e-07,
"loss": 0.3313198387622833,
"step": 1388
},
{
"epoch": 1.9027397260273973,
"grad_norm": 0.16718758642673492,
"learning_rate": 1.2731645278655448e-07,
"loss": 0.3705058991909027,
"step": 1389
},
{
"epoch": 1.904109589041096,
"grad_norm": 0.16766990721225739,
"learning_rate": 1.2381170954955611e-07,
"loss": 0.3541448712348938,
"step": 1390
},
{
"epoch": 1.9054794520547946,
"grad_norm": 0.16111250221729279,
"learning_rate": 1.203555804317813e-07,
"loss": 0.31774333119392395,
"step": 1391
},
{
"epoch": 1.9068493150684933,
"grad_norm": 0.15901249647140503,
"learning_rate": 1.1694808244552824e-07,
"loss": 0.32911622524261475,
"step": 1392
},
{
"epoch": 1.9082191780821918,
"grad_norm": 0.15923842787742615,
"learning_rate": 1.1358923236371888e-07,
"loss": 0.35313525795936584,
"step": 1393
},
{
"epoch": 1.9095890410958904,
"grad_norm": 0.15409809350967407,
"learning_rate": 1.1027904671981004e-07,
"loss": 0.3332017660140991,
"step": 1394
},
{
"epoch": 1.910958904109589,
"grad_norm": 0.162120521068573,
"learning_rate": 1.0701754180771462e-07,
"loss": 0.35140013694763184,
"step": 1395
},
{
"epoch": 1.9123287671232876,
"grad_norm": 0.15148550271987915,
"learning_rate": 1.0380473368172272e-07,
"loss": 0.3212718963623047,
"step": 1396
},
{
"epoch": 1.9136986301369863,
"grad_norm": 0.1591331958770752,
"learning_rate": 1.0064063815642178e-07,
"loss": 0.3234368860721588,
"step": 1397
},
{
"epoch": 1.915068493150685,
"grad_norm": 0.15532776713371277,
"learning_rate": 9.752527080661655e-08,
"loss": 0.36444076895713806,
"step": 1398
},
{
"epoch": 1.9164383561643836,
"grad_norm": 0.13341137766838074,
"learning_rate": 9.445864696725704e-08,
"loss": 0.29276254773139954,
"step": 1399
},
{
"epoch": 1.9178082191780823,
"grad_norm": 0.15290893614292145,
"learning_rate": 9.144078173335847e-08,
"loss": 0.32714784145355225,
"step": 1400
},
{
"epoch": 1.919178082191781,
"grad_norm": 0.15745221078395844,
"learning_rate": 8.847168995992916e-08,
"loss": 0.3503940999507904,
"step": 1401
},
{
"epoch": 1.9205479452054794,
"grad_norm": 0.17177851498126984,
"learning_rate": 8.555138626189619e-08,
"loss": 0.3281104862689972,
"step": 1402
},
{
"epoch": 1.9219178082191781,
"grad_norm": 0.14865048229694366,
"learning_rate": 8.267988501403645e-08,
"loss": 0.328127920627594,
"step": 1403
},
{
"epoch": 1.9232876712328766,
"grad_norm": 0.15570548176765442,
"learning_rate": 7.985720035090239e-08,
"loss": 0.3238529860973358,
"step": 1404
},
{
"epoch": 1.9246575342465753,
"grad_norm": 0.1401631385087967,
"learning_rate": 7.708334616675417e-08,
"loss": 0.29453912377357483,
"step": 1405
},
{
"epoch": 1.926027397260274,
"grad_norm": 0.15698212385177612,
"learning_rate": 7.435833611549204e-08,
"loss": 0.3361373841762543,
"step": 1406
},
{
"epoch": 1.9273972602739726,
"grad_norm": 0.1749168485403061,
"learning_rate": 7.16821836105841e-08,
"loss": 0.3558928668498993,
"step": 1407
},
{
"epoch": 1.9287671232876713,
"grad_norm": 0.16780687868595123,
"learning_rate": 6.905490182500863e-08,
"loss": 0.3293951749801636,
"step": 1408
},
{
"epoch": 1.93013698630137,
"grad_norm": 0.15367259085178375,
"learning_rate": 6.647650369118408e-08,
"loss": 0.29588863253593445,
"step": 1409
},
{
"epoch": 1.9315068493150684,
"grad_norm": 0.15608543157577515,
"learning_rate": 6.394700190090252e-08,
"loss": 0.33914613723754883,
"step": 1410
},
{
"epoch": 1.9328767123287671,
"grad_norm": 0.17230366170406342,
"learning_rate": 6.146640890527634e-08,
"loss": 0.36092859506607056,
"step": 1411
},
{
"epoch": 1.9342465753424658,
"grad_norm": 0.16014167666435242,
"learning_rate": 5.903473691466599e-08,
"loss": 0.3141736686229706,
"step": 1412
},
{
"epoch": 1.9356164383561643,
"grad_norm": 0.17919820547103882,
"learning_rate": 5.665199789862907e-08,
"loss": 0.38164132833480835,
"step": 1413
},
{
"epoch": 1.936986301369863,
"grad_norm": 0.14225740730762482,
"learning_rate": 5.431820358585582e-08,
"loss": 0.333599328994751,
"step": 1414
},
{
"epoch": 1.9383561643835616,
"grad_norm": 0.16404688358306885,
"learning_rate": 5.203336546411475e-08,
"loss": 0.3227609395980835,
"step": 1415
},
{
"epoch": 1.9397260273972603,
"grad_norm": 0.15278957784175873,
"learning_rate": 4.979749478019158e-08,
"loss": 0.3225298523902893,
"step": 1416
},
{
"epoch": 1.941095890410959,
"grad_norm": 0.16982510685920715,
"learning_rate": 4.761060253984151e-08,
"loss": 0.3580218553543091,
"step": 1417
},
{
"epoch": 1.9424657534246577,
"grad_norm": 0.15608397126197815,
"learning_rate": 4.547269950772593e-08,
"loss": 0.3315258324146271,
"step": 1418
},
{
"epoch": 1.9438356164383561,
"grad_norm": 0.15200930833816528,
"learning_rate": 4.338379620736577e-08,
"loss": 0.34367135167121887,
"step": 1419
},
{
"epoch": 1.9452054794520548,
"grad_norm": 0.163382887840271,
"learning_rate": 4.1343902921088255e-08,
"loss": 0.3307231664657593,
"step": 1420
},
{
"epoch": 1.9465753424657535,
"grad_norm": 0.15526485443115234,
"learning_rate": 3.9353029689974676e-08,
"loss": 0.3520579934120178,
"step": 1421
},
{
"epoch": 1.947945205479452,
"grad_norm": 0.16491355001926422,
"learning_rate": 3.741118631381269e-08,
"loss": 0.3547651767730713,
"step": 1422
},
{
"epoch": 1.9493150684931506,
"grad_norm": 0.16033078730106354,
"learning_rate": 3.551838235104854e-08,
"loss": 0.34746053814888,
"step": 1423
},
{
"epoch": 1.9506849315068493,
"grad_norm": 0.15535302460193634,
"learning_rate": 3.3674627118737144e-08,
"loss": 0.3341052830219269,
"step": 1424
},
{
"epoch": 1.952054794520548,
"grad_norm": 0.16157664358615875,
"learning_rate": 3.187992969249876e-08,
"loss": 0.3238033652305603,
"step": 1425
},
{
"epoch": 1.9534246575342467,
"grad_norm": 0.16438177227973938,
"learning_rate": 3.013429890647457e-08,
"loss": 0.3472440242767334,
"step": 1426
},
{
"epoch": 1.9547945205479453,
"grad_norm": 0.16162332892417908,
"learning_rate": 2.84377433532812e-08,
"loss": 0.3489248752593994,
"step": 1427
},
{
"epoch": 1.9561643835616438,
"grad_norm": 0.1682146191596985,
"learning_rate": 2.6790271383970723e-08,
"loss": 0.3327878713607788,
"step": 1428
},
{
"epoch": 1.9575342465753425,
"grad_norm": 0.1820368468761444,
"learning_rate": 2.5191891107985143e-08,
"loss": 0.37263011932373047,
"step": 1429
},
{
"epoch": 1.958904109589041,
"grad_norm": 0.14847320318222046,
"learning_rate": 2.364261039312421e-08,
"loss": 0.3151990473270416,
"step": 1430
},
{
"epoch": 1.9602739726027396,
"grad_norm": 0.16611696779727936,
"learning_rate": 2.2142436865499884e-08,
"loss": 0.34913647174835205,
"step": 1431
},
{
"epoch": 1.9616438356164383,
"grad_norm": 0.15534548461437225,
"learning_rate": 2.06913779094986e-08,
"loss": 0.3297916054725647,
"step": 1432
},
{
"epoch": 1.963013698630137,
"grad_norm": 0.148855060338974,
"learning_rate": 1.92894406677524e-08,
"loss": 0.2957698702812195,
"step": 1433
},
{
"epoch": 1.9643835616438357,
"grad_norm": 0.1566469967365265,
"learning_rate": 1.7936632041094527e-08,
"loss": 0.3352767825126648,
"step": 1434
},
{
"epoch": 1.9657534246575343,
"grad_norm": 0.16014723479747772,
"learning_rate": 1.6632958688532765e-08,
"loss": 0.35956257581710815,
"step": 1435
},
{
"epoch": 1.967123287671233,
"grad_norm": 0.16210445761680603,
"learning_rate": 1.5378427027210597e-08,
"loss": 0.33658331632614136,
"step": 1436
},
{
"epoch": 1.9684931506849315,
"grad_norm": 0.1651066094636917,
"learning_rate": 1.4173043232380557e-08,
"loss": 0.35282987356185913,
"step": 1437
},
{
"epoch": 1.9698630136986301,
"grad_norm": 0.15467463433742523,
"learning_rate": 1.3016813237372028e-08,
"loss": 0.3518558144569397,
"step": 1438
},
{
"epoch": 1.9712328767123286,
"grad_norm": 0.14918740093708038,
"learning_rate": 1.190974273355905e-08,
"loss": 0.3505173921585083,
"step": 1439
},
{
"epoch": 1.9726027397260273,
"grad_norm": 0.14169228076934814,
"learning_rate": 1.0851837170340329e-08,
"loss": 0.29582536220550537,
"step": 1440
},
{
"epoch": 1.973972602739726,
"grad_norm": 0.16460123658180237,
"learning_rate": 9.843101755101503e-09,
"loss": 0.31888025999069214,
"step": 1441
},
{
"epoch": 1.9753424657534246,
"grad_norm": 0.15857937932014465,
"learning_rate": 8.883541453199585e-09,
"loss": 0.31724226474761963,
"step": 1442
},
{
"epoch": 1.9767123287671233,
"grad_norm": 0.15296325087547302,
"learning_rate": 7.973160987931883e-09,
"loss": 0.33146071434020996,
"step": 1443
},
{
"epoch": 1.978082191780822,
"grad_norm": 0.15920867025852203,
"learning_rate": 7.111964840514907e-09,
"loss": 0.33872342109680176,
"step": 1444
},
{
"epoch": 1.9794520547945207,
"grad_norm": 0.16633795201778412,
"learning_rate": 6.299957250064381e-09,
"loss": 0.33016908168792725,
"step": 1445
},
{
"epoch": 1.9808219178082191,
"grad_norm": 0.15383541584014893,
"learning_rate": 5.537142213569713e-09,
"loss": 0.3443043529987335,
"step": 1446
},
{
"epoch": 1.9821917808219178,
"grad_norm": 0.15127472579479218,
"learning_rate": 4.823523485879556e-09,
"loss": 0.3027114272117615,
"step": 1447
},
{
"epoch": 1.9835616438356163,
"grad_norm": 0.15184855461120605,
"learning_rate": 4.159104579678497e-09,
"loss": 0.34180566668510437,
"step": 1448
},
{
"epoch": 1.984931506849315,
"grad_norm": 0.14583970606327057,
"learning_rate": 3.5438887654737355e-09,
"loss": 0.31200599670410156,
"step": 1449
},
{
"epoch": 1.9863013698630136,
"grad_norm": 0.16080637276172638,
"learning_rate": 2.9778790715784267e-09,
"loss": 0.3215121924877167,
"step": 1450
},
{
"epoch": 1.9876712328767123,
"grad_norm": 0.16216839849948883,
"learning_rate": 2.4610782840917003e-09,
"loss": 0.3597952127456665,
"step": 1451
},
{
"epoch": 1.989041095890411,
"grad_norm": 0.15053534507751465,
"learning_rate": 1.993488946891997e-09,
"loss": 0.32388773560523987,
"step": 1452
},
{
"epoch": 1.9904109589041097,
"grad_norm": 0.15643882751464844,
"learning_rate": 1.5751133616215276e-09,
"loss": 0.31341126561164856,
"step": 1453
},
{
"epoch": 1.9917808219178084,
"grad_norm": 0.14695680141448975,
"learning_rate": 1.2059535876729478e-09,
"loss": 0.3207545280456543,
"step": 1454
},
{
"epoch": 1.9931506849315068,
"grad_norm": 0.15658403933048248,
"learning_rate": 8.860114421826993e-10,
"loss": 0.30730488896369934,
"step": 1455
},
{
"epoch": 1.9945205479452055,
"grad_norm": 0.14001530408859253,
"learning_rate": 6.15288500019906e-10,
"loss": 0.31398269534111023,
"step": 1456
},
{
"epoch": 1.995890410958904,
"grad_norm": 0.14669930934906006,
"learning_rate": 3.9378609377971335e-10,
"loss": 0.33043134212493896,
"step": 1457
},
{
"epoch": 1.9972602739726026,
"grad_norm": 0.1539190709590912,
"learning_rate": 2.2150531377551633e-10,
"loss": 0.30559971928596497,
"step": 1458
},
{
"epoch": 1.9986301369863013,
"grad_norm": 0.1698891520500183,
"learning_rate": 9.844700803562923e-11,
"loss": 0.35297560691833496,
"step": 1459
},
{
"epoch": 2.0,
"grad_norm": 0.164133220911026,
"learning_rate": 2.4611782295513864e-11,
"loss": 0.3423645496368408,
"step": 1460
}
],
"logging_steps": 1,
"max_steps": 1460,
"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": 1.5316438499721216e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}