ocr-tr-2 / trainer_state.json
Ba2han's picture
Upload folder using huggingface_hub
5480203 verified
Raw
History Blame Contribute Delete
191 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.2469437652811737,
"eval_steps": 500,
"global_step": 1020,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0012224938875305623,
"grad_norm": 1.044545292854309,
"learning_rate": 2e-06,
"loss": 0.16098472476005554,
"step": 1
},
{
"epoch": 0.0024449877750611247,
"grad_norm": 1.3195806741714478,
"learning_rate": 1.9999981562458212e-06,
"loss": 0.17797882854938507,
"step": 2
},
{
"epoch": 0.003667481662591687,
"grad_norm": 0.8549019694328308,
"learning_rate": 1.9999926249900844e-06,
"loss": 0.09693044424057007,
"step": 3
},
{
"epoch": 0.004889975550122249,
"grad_norm": 1.310805082321167,
"learning_rate": 1.9999834062531854e-06,
"loss": 0.1697659194469452,
"step": 4
},
{
"epoch": 0.006112469437652812,
"grad_norm": 1.06551992893219,
"learning_rate": 1.9999705000691188e-06,
"loss": 0.12446862459182739,
"step": 5
},
{
"epoch": 0.007334963325183374,
"grad_norm": 4.9082207679748535,
"learning_rate": 1.999953906485476e-06,
"loss": 0.31172120571136475,
"step": 6
},
{
"epoch": 0.008557457212713936,
"grad_norm": 1.3029686212539673,
"learning_rate": 1.9999336255634463e-06,
"loss": 0.1647353172302246,
"step": 7
},
{
"epoch": 0.009779951100244499,
"grad_norm": 1.0400989055633545,
"learning_rate": 1.9999096573778155e-06,
"loss": 0.09065035730600357,
"step": 8
},
{
"epoch": 0.011002444987775062,
"grad_norm": 0.8220394253730774,
"learning_rate": 1.9998820020169668e-06,
"loss": 0.06520912796258926,
"step": 9
},
{
"epoch": 0.012224938875305624,
"grad_norm": 0.909023642539978,
"learning_rate": 1.999850659582879e-06,
"loss": 0.08835319429636002,
"step": 10
},
{
"epoch": 0.013447432762836185,
"grad_norm": 0.9066786766052246,
"learning_rate": 1.9998156301911284e-06,
"loss": 0.06585387140512466,
"step": 11
},
{
"epoch": 0.014669926650366748,
"grad_norm": 0.9607728719711304,
"learning_rate": 1.999776913970885e-06,
"loss": 0.0931173786520958,
"step": 12
},
{
"epoch": 0.01589242053789731,
"grad_norm": 5.249514102935791,
"learning_rate": 1.9997345110649166e-06,
"loss": 0.2670130133628845,
"step": 13
},
{
"epoch": 0.017114914425427872,
"grad_norm": 0.9609458446502686,
"learning_rate": 1.9996884216295834e-06,
"loss": 0.06627589464187622,
"step": 14
},
{
"epoch": 0.018337408312958436,
"grad_norm": 0.9252450466156006,
"learning_rate": 1.999638645834841e-06,
"loss": 0.07454587519168854,
"step": 15
},
{
"epoch": 0.019559902200488997,
"grad_norm": 1.022990345954895,
"learning_rate": 1.9995851838642375e-06,
"loss": 0.08483454585075378,
"step": 16
},
{
"epoch": 0.02078239608801956,
"grad_norm": 1.0671887397766113,
"learning_rate": 1.999528035914915e-06,
"loss": 0.10120941698551178,
"step": 17
},
{
"epoch": 0.022004889975550123,
"grad_norm": 0.8922954797744751,
"learning_rate": 1.9994672021976063e-06,
"loss": 0.07356676459312439,
"step": 18
},
{
"epoch": 0.023227383863080684,
"grad_norm": 1.2407742738723755,
"learning_rate": 1.999402682936637e-06,
"loss": 0.14839434623718262,
"step": 19
},
{
"epoch": 0.02444987775061125,
"grad_norm": 0.6839045882225037,
"learning_rate": 1.9993344783699217e-06,
"loss": 0.06802748888731003,
"step": 20
},
{
"epoch": 0.02567237163814181,
"grad_norm": 0.986915647983551,
"learning_rate": 1.999262588748966e-06,
"loss": 0.08436816185712814,
"step": 21
},
{
"epoch": 0.02689486552567237,
"grad_norm": 1.240136981010437,
"learning_rate": 1.999187014338863e-06,
"loss": 0.16614878177642822,
"step": 22
},
{
"epoch": 0.028117359413202935,
"grad_norm": 5.780612468719482,
"learning_rate": 1.9991077554182944e-06,
"loss": 0.25963252782821655,
"step": 23
},
{
"epoch": 0.029339853300733496,
"grad_norm": 0.9097835421562195,
"learning_rate": 1.9990248122795273e-06,
"loss": 0.08546145260334015,
"step": 24
},
{
"epoch": 0.030562347188264057,
"grad_norm": 1.2009776830673218,
"learning_rate": 1.9989381852284164e-06,
"loss": 0.16851800680160522,
"step": 25
},
{
"epoch": 0.03178484107579462,
"grad_norm": 1.3855910301208496,
"learning_rate": 1.998847874584399e-06,
"loss": 0.19520896673202515,
"step": 26
},
{
"epoch": 0.03300733496332518,
"grad_norm": 0.9027628898620605,
"learning_rate": 1.9987538806804963e-06,
"loss": 0.0939425379037857,
"step": 27
},
{
"epoch": 0.034229828850855744,
"grad_norm": 1.0047492980957031,
"learning_rate": 1.9986562038633113e-06,
"loss": 0.09135263413190842,
"step": 28
},
{
"epoch": 0.035452322738386305,
"grad_norm": 0.7804238796234131,
"learning_rate": 1.998554844493029e-06,
"loss": 0.06966346502304077,
"step": 29
},
{
"epoch": 0.03667481662591687,
"grad_norm": 5.010461330413818,
"learning_rate": 1.9984498029434124e-06,
"loss": 0.28327855467796326,
"step": 30
},
{
"epoch": 0.037897310513447434,
"grad_norm": 0.789050281047821,
"learning_rate": 1.998341079601803e-06,
"loss": 0.08065496385097504,
"step": 31
},
{
"epoch": 0.039119804400977995,
"grad_norm": 1.1024219989776611,
"learning_rate": 1.998228674869119e-06,
"loss": 0.09279895573854446,
"step": 32
},
{
"epoch": 0.040342298288508556,
"grad_norm": 0.6993586421012878,
"learning_rate": 1.9981125891598544e-06,
"loss": 0.048593733459711075,
"step": 33
},
{
"epoch": 0.04156479217603912,
"grad_norm": 0.9050285816192627,
"learning_rate": 1.9979928229020755e-06,
"loss": 0.07057932019233704,
"step": 34
},
{
"epoch": 0.042787286063569685,
"grad_norm": 4.215475559234619,
"learning_rate": 1.9978693765374215e-06,
"loss": 0.2667696177959442,
"step": 35
},
{
"epoch": 0.044009779951100246,
"grad_norm": 1.0129759311676025,
"learning_rate": 1.997742250521102e-06,
"loss": 0.09877517074346542,
"step": 36
},
{
"epoch": 0.04523227383863081,
"grad_norm": 2.5444610118865967,
"learning_rate": 1.997611445321896e-06,
"loss": 0.17668884992599487,
"step": 37
},
{
"epoch": 0.04645476772616137,
"grad_norm": 0.8873361349105835,
"learning_rate": 1.9974769614221474e-06,
"loss": 0.05411791056394577,
"step": 38
},
{
"epoch": 0.04767726161369193,
"grad_norm": 0.8752514123916626,
"learning_rate": 1.997338799317767e-06,
"loss": 0.09558999538421631,
"step": 39
},
{
"epoch": 0.0488997555012225,
"grad_norm": 0.6304615139961243,
"learning_rate": 1.99719695951823e-06,
"loss": 0.06795249134302139,
"step": 40
},
{
"epoch": 0.05012224938875306,
"grad_norm": 0.8881245851516724,
"learning_rate": 1.99705144254657e-06,
"loss": 0.07396651804447174,
"step": 41
},
{
"epoch": 0.05134474327628362,
"grad_norm": 0.7423112988471985,
"learning_rate": 1.9969022489393837e-06,
"loss": 0.08127471804618835,
"step": 42
},
{
"epoch": 0.05256723716381418,
"grad_norm": 0.6915601491928101,
"learning_rate": 1.996749379246823e-06,
"loss": 0.06632934510707855,
"step": 43
},
{
"epoch": 0.05378973105134474,
"grad_norm": 0.8911988735198975,
"learning_rate": 1.9965928340325957e-06,
"loss": 0.11288589239120483,
"step": 44
},
{
"epoch": 0.0550122249388753,
"grad_norm": 0.6879037022590637,
"learning_rate": 1.9964326138739645e-06,
"loss": 0.07341636717319489,
"step": 45
},
{
"epoch": 0.05623471882640587,
"grad_norm": 1.097665548324585,
"learning_rate": 1.996268719361742e-06,
"loss": 0.10401637852191925,
"step": 46
},
{
"epoch": 0.05745721271393643,
"grad_norm": 0.9345325827598572,
"learning_rate": 1.9961011511002907e-06,
"loss": 0.06859036535024643,
"step": 47
},
{
"epoch": 0.05867970660146699,
"grad_norm": 0.9359571933746338,
"learning_rate": 1.99592990970752e-06,
"loss": 0.09040691703557968,
"step": 48
},
{
"epoch": 0.05990220048899755,
"grad_norm": 0.7405828237533569,
"learning_rate": 1.995754995814884e-06,
"loss": 0.08165255188941956,
"step": 49
},
{
"epoch": 0.061124694376528114,
"grad_norm": 1.1201773881912231,
"learning_rate": 1.995576410067379e-06,
"loss": 0.10577505826950073,
"step": 50
},
{
"epoch": 0.06234718826405868,
"grad_norm": 0.766424834728241,
"learning_rate": 1.995394153123542e-06,
"loss": 0.051467303186655045,
"step": 51
},
{
"epoch": 0.06356968215158924,
"grad_norm": 0.8485493063926697,
"learning_rate": 1.9952082256554464e-06,
"loss": 0.10404057055711746,
"step": 52
},
{
"epoch": 0.0647921760391198,
"grad_norm": 0.9804834723472595,
"learning_rate": 1.9950186283487016e-06,
"loss": 0.12650075554847717,
"step": 53
},
{
"epoch": 0.06601466992665037,
"grad_norm": 0.8129968047142029,
"learning_rate": 1.9948253619024487e-06,
"loss": 0.11533457040786743,
"step": 54
},
{
"epoch": 0.06723716381418093,
"grad_norm": 0.8469319343566895,
"learning_rate": 1.99462842702936e-06,
"loss": 0.0810215175151825,
"step": 55
},
{
"epoch": 0.06845965770171149,
"grad_norm": 0.6489520072937012,
"learning_rate": 1.9944278244556343e-06,
"loss": 0.06589583307504654,
"step": 56
},
{
"epoch": 0.06968215158924206,
"grad_norm": 0.7432178258895874,
"learning_rate": 1.994223554920995e-06,
"loss": 0.11960537731647491,
"step": 57
},
{
"epoch": 0.07090464547677261,
"grad_norm": 0.5680833458900452,
"learning_rate": 1.9940156191786883e-06,
"loss": 0.057387810200452805,
"step": 58
},
{
"epoch": 0.07212713936430318,
"grad_norm": 0.7835686206817627,
"learning_rate": 1.9938040179954784e-06,
"loss": 0.07761581242084503,
"step": 59
},
{
"epoch": 0.07334963325183375,
"grad_norm": 0.8264024257659912,
"learning_rate": 1.9935887521516466e-06,
"loss": 0.06637915968894958,
"step": 60
},
{
"epoch": 0.0745721271393643,
"grad_norm": 0.8332459926605225,
"learning_rate": 1.9933698224409872e-06,
"loss": 0.10251758992671967,
"step": 61
},
{
"epoch": 0.07579462102689487,
"grad_norm": 0.8895119428634644,
"learning_rate": 1.993147229670806e-06,
"loss": 0.08672869950532913,
"step": 62
},
{
"epoch": 0.07701711491442542,
"grad_norm": 3.7527127265930176,
"learning_rate": 1.9929209746619157e-06,
"loss": 0.21330073475837708,
"step": 63
},
{
"epoch": 0.07823960880195599,
"grad_norm": 0.8739458322525024,
"learning_rate": 1.9926910582486325e-06,
"loss": 0.15269967913627625,
"step": 64
},
{
"epoch": 0.07946210268948656,
"grad_norm": 0.6081534624099731,
"learning_rate": 1.9924574812787766e-06,
"loss": 0.054524220526218414,
"step": 65
},
{
"epoch": 0.08068459657701711,
"grad_norm": 1.0636615753173828,
"learning_rate": 1.992220244613664e-06,
"loss": 0.10620469599962234,
"step": 66
},
{
"epoch": 0.08190709046454768,
"grad_norm": 0.8420392870903015,
"learning_rate": 1.9919793491281066e-06,
"loss": 0.07811391353607178,
"step": 67
},
{
"epoch": 0.08312958435207823,
"grad_norm": 0.6580442190170288,
"learning_rate": 1.9917347957104095e-06,
"loss": 0.05853688716888428,
"step": 68
},
{
"epoch": 0.0843520782396088,
"grad_norm": 0.6775596141815186,
"learning_rate": 1.991486585262365e-06,
"loss": 0.0882895290851593,
"step": 69
},
{
"epoch": 0.08557457212713937,
"grad_norm": 5.316723346710205,
"learning_rate": 1.991234718699251e-06,
"loss": 0.2285734862089157,
"step": 70
},
{
"epoch": 0.08679706601466992,
"grad_norm": 0.7672296166419983,
"learning_rate": 1.9909791969498276e-06,
"loss": 0.061234671622514725,
"step": 71
},
{
"epoch": 0.08801955990220049,
"grad_norm": 0.5419216156005859,
"learning_rate": 1.990720020956334e-06,
"loss": 0.03853198140859604,
"step": 72
},
{
"epoch": 0.08924205378973105,
"grad_norm": 0.8963152766227722,
"learning_rate": 1.9904571916744835e-06,
"loss": 0.11935413628816605,
"step": 73
},
{
"epoch": 0.09046454767726161,
"grad_norm": 0.49984046816825867,
"learning_rate": 1.9901907100734606e-06,
"loss": 0.05078790709376335,
"step": 74
},
{
"epoch": 0.09168704156479218,
"grad_norm": 0.7401895523071289,
"learning_rate": 1.9899205771359193e-06,
"loss": 0.10604439675807953,
"step": 75
},
{
"epoch": 0.09290953545232274,
"grad_norm": 0.7164925336837769,
"learning_rate": 1.9896467938579772e-06,
"loss": 0.037979379296302795,
"step": 76
},
{
"epoch": 0.0941320293398533,
"grad_norm": 0.6726301312446594,
"learning_rate": 1.989369361249211e-06,
"loss": 0.06786077469587326,
"step": 77
},
{
"epoch": 0.09535452322738386,
"grad_norm": 0.9190412759780884,
"learning_rate": 1.9890882803326578e-06,
"loss": 0.12730944156646729,
"step": 78
},
{
"epoch": 0.09657701711491443,
"grad_norm": 0.7142415642738342,
"learning_rate": 1.988803552144804e-06,
"loss": 0.049502529203891754,
"step": 79
},
{
"epoch": 0.097799511002445,
"grad_norm": 0.6762092709541321,
"learning_rate": 1.9885151777355884e-06,
"loss": 0.06278107315301895,
"step": 80
},
{
"epoch": 0.09902200488997555,
"grad_norm": 0.8819540143013,
"learning_rate": 1.9882231581683933e-06,
"loss": 0.11831459403038025,
"step": 81
},
{
"epoch": 0.10024449877750612,
"grad_norm": 0.8594198226928711,
"learning_rate": 1.987927494520044e-06,
"loss": 0.07652519643306732,
"step": 82
},
{
"epoch": 0.10146699266503667,
"grad_norm": 0.4754820466041565,
"learning_rate": 1.9876281878808017e-06,
"loss": 0.03188315033912659,
"step": 83
},
{
"epoch": 0.10268948655256724,
"grad_norm": 0.6299994587898254,
"learning_rate": 1.987325239354363e-06,
"loss": 0.05682877451181412,
"step": 84
},
{
"epoch": 0.1039119804400978,
"grad_norm": 0.5381066799163818,
"learning_rate": 1.987018650057853e-06,
"loss": 0.05135112255811691,
"step": 85
},
{
"epoch": 0.10513447432762836,
"grad_norm": 3.665609836578369,
"learning_rate": 1.986708421121822e-06,
"loss": 0.24795880913734436,
"step": 86
},
{
"epoch": 0.10635696821515893,
"grad_norm": 0.5929223895072937,
"learning_rate": 1.986394553690242e-06,
"loss": 0.05152888968586922,
"step": 87
},
{
"epoch": 0.10757946210268948,
"grad_norm": 0.7831748723983765,
"learning_rate": 1.9860770489205013e-06,
"loss": 0.07349374145269394,
"step": 88
},
{
"epoch": 0.10880195599022005,
"grad_norm": 0.6961215138435364,
"learning_rate": 1.9857559079834022e-06,
"loss": 0.06047668680548668,
"step": 89
},
{
"epoch": 0.1100244498777506,
"grad_norm": 0.6508141756057739,
"learning_rate": 1.985431132063154e-06,
"loss": 0.060005564242601395,
"step": 90
},
{
"epoch": 0.11124694376528117,
"grad_norm": 0.816543459892273,
"learning_rate": 1.9851027223573706e-06,
"loss": 0.06605978310108185,
"step": 91
},
{
"epoch": 0.11246943765281174,
"grad_norm": 0.6350448727607727,
"learning_rate": 1.984770680077066e-06,
"loss": 0.05193399637937546,
"step": 92
},
{
"epoch": 0.1136919315403423,
"grad_norm": 0.5598679780960083,
"learning_rate": 1.9844350064466482e-06,
"loss": 0.0424959622323513,
"step": 93
},
{
"epoch": 0.11491442542787286,
"grad_norm": 0.8493036031723022,
"learning_rate": 1.9840957027039175e-06,
"loss": 0.1120591014623642,
"step": 94
},
{
"epoch": 0.11613691931540342,
"grad_norm": 0.9302883148193359,
"learning_rate": 1.9837527701000587e-06,
"loss": 0.1210164949297905,
"step": 95
},
{
"epoch": 0.11735941320293398,
"grad_norm": 0.6914235353469849,
"learning_rate": 1.9834062098996383e-06,
"loss": 0.07304820418357849,
"step": 96
},
{
"epoch": 0.11858190709046455,
"grad_norm": 0.545456051826477,
"learning_rate": 1.9830560233806002e-06,
"loss": 0.05308880656957626,
"step": 97
},
{
"epoch": 0.1198044009779951,
"grad_norm": 0.9245759844779968,
"learning_rate": 1.9827022118342603e-06,
"loss": 0.12079302221536636,
"step": 98
},
{
"epoch": 0.12102689486552567,
"grad_norm": 0.7079183459281921,
"learning_rate": 1.982344776565302e-06,
"loss": 0.08940428495407104,
"step": 99
},
{
"epoch": 0.12224938875305623,
"grad_norm": 0.6680834293365479,
"learning_rate": 1.98198371889177e-06,
"loss": 0.05445454269647598,
"step": 100
},
{
"epoch": 0.1234718826405868,
"grad_norm": 1.16506028175354,
"learning_rate": 1.9816190401450678e-06,
"loss": 0.09130517393350601,
"step": 101
},
{
"epoch": 0.12469437652811736,
"grad_norm": 0.7334263920783997,
"learning_rate": 1.9812507416699517e-06,
"loss": 0.05265149474143982,
"step": 102
},
{
"epoch": 0.12591687041564792,
"grad_norm": 0.4865010380744934,
"learning_rate": 1.9808788248245245e-06,
"loss": 0.058592330664396286,
"step": 103
},
{
"epoch": 0.1271393643031785,
"grad_norm": 0.7385439276695251,
"learning_rate": 1.9805032909802338e-06,
"loss": 0.08480134606361389,
"step": 104
},
{
"epoch": 0.12836185819070906,
"grad_norm": 0.7237193584442139,
"learning_rate": 1.9801241415218635e-06,
"loss": 0.07208502292633057,
"step": 105
},
{
"epoch": 0.1295843520782396,
"grad_norm": 0.7185537219047546,
"learning_rate": 1.9797413778475297e-06,
"loss": 0.05598007142543793,
"step": 106
},
{
"epoch": 0.13080684596577016,
"grad_norm": 0.56794673204422,
"learning_rate": 1.9793550013686772e-06,
"loss": 0.051465995609760284,
"step": 107
},
{
"epoch": 0.13202933985330073,
"grad_norm": 1.1163580417633057,
"learning_rate": 1.9789650135100723e-06,
"loss": 0.13626353442668915,
"step": 108
},
{
"epoch": 0.1332518337408313,
"grad_norm": 0.8450824022293091,
"learning_rate": 1.978571415709799e-06,
"loss": 0.11162219196557999,
"step": 109
},
{
"epoch": 0.13447432762836187,
"grad_norm": 0.9856062531471252,
"learning_rate": 1.9781742094192516e-06,
"loss": 0.12988917529582977,
"step": 110
},
{
"epoch": 0.1356968215158924,
"grad_norm": 0.6862197518348694,
"learning_rate": 1.9777733961031326e-06,
"loss": 0.046755600720644,
"step": 111
},
{
"epoch": 0.13691931540342298,
"grad_norm": 0.8343086838722229,
"learning_rate": 1.9773689772394434e-06,
"loss": 0.06609448045492172,
"step": 112
},
{
"epoch": 0.13814180929095354,
"grad_norm": 4.203027725219727,
"learning_rate": 1.9769609543194825e-06,
"loss": 0.23921093344688416,
"step": 113
},
{
"epoch": 0.1393643031784841,
"grad_norm": 0.805291473865509,
"learning_rate": 1.976549328847838e-06,
"loss": 0.10714933276176453,
"step": 114
},
{
"epoch": 0.14058679706601468,
"grad_norm": 0.6103475689888,
"learning_rate": 1.9761341023423824e-06,
"loss": 0.06422054767608643,
"step": 115
},
{
"epoch": 0.14180929095354522,
"grad_norm": 0.6594969630241394,
"learning_rate": 1.9757152763342662e-06,
"loss": 0.053227499127388,
"step": 116
},
{
"epoch": 0.1430317848410758,
"grad_norm": 0.6954916715621948,
"learning_rate": 1.975292852367914e-06,
"loss": 0.06985656917095184,
"step": 117
},
{
"epoch": 0.14425427872860636,
"grad_norm": 0.5022503733634949,
"learning_rate": 1.974866832001018e-06,
"loss": 0.041988302022218704,
"step": 118
},
{
"epoch": 0.14547677261613692,
"grad_norm": 0.7095266580581665,
"learning_rate": 1.9744372168045322e-06,
"loss": 0.06460104137659073,
"step": 119
},
{
"epoch": 0.1466992665036675,
"grad_norm": 3.1877522468566895,
"learning_rate": 1.9740040083626654e-06,
"loss": 0.15561164915561676,
"step": 120
},
{
"epoch": 0.14792176039119803,
"grad_norm": 1.154474139213562,
"learning_rate": 1.973567208272878e-06,
"loss": 0.11057989299297333,
"step": 121
},
{
"epoch": 0.1491442542787286,
"grad_norm": 0.6467365026473999,
"learning_rate": 1.9731268181458734e-06,
"loss": 0.05472421273589134,
"step": 122
},
{
"epoch": 0.15036674816625917,
"grad_norm": 0.8463435173034668,
"learning_rate": 1.9726828396055946e-06,
"loss": 0.0851202979683876,
"step": 123
},
{
"epoch": 0.15158924205378974,
"grad_norm": 0.7386101484298706,
"learning_rate": 1.9722352742892154e-06,
"loss": 0.053707897663116455,
"step": 124
},
{
"epoch": 0.1528117359413203,
"grad_norm": 0.9296069741249084,
"learning_rate": 1.971784123847137e-06,
"loss": 0.09668777883052826,
"step": 125
},
{
"epoch": 0.15403422982885084,
"grad_norm": 0.6150487661361694,
"learning_rate": 1.971329389942981e-06,
"loss": 0.06070238724350929,
"step": 126
},
{
"epoch": 0.1552567237163814,
"grad_norm": 0.5244233012199402,
"learning_rate": 1.9708710742535813e-06,
"loss": 0.06360621750354767,
"step": 127
},
{
"epoch": 0.15647921760391198,
"grad_norm": 0.5770973563194275,
"learning_rate": 1.9704091784689817e-06,
"loss": 0.0655631273984909,
"step": 128
},
{
"epoch": 0.15770171149144255,
"grad_norm": 0.6751277446746826,
"learning_rate": 1.9699437042924264e-06,
"loss": 0.089437335729599,
"step": 129
},
{
"epoch": 0.15892420537897312,
"grad_norm": 0.8923412561416626,
"learning_rate": 1.9694746534403553e-06,
"loss": 0.10819036513566971,
"step": 130
},
{
"epoch": 0.16014669926650366,
"grad_norm": 0.6662521958351135,
"learning_rate": 1.969002027642398e-06,
"loss": 0.04933347925543785,
"step": 131
},
{
"epoch": 0.16136919315403422,
"grad_norm": 0.6493452787399292,
"learning_rate": 1.968525828641365e-06,
"loss": 0.0524442158639431,
"step": 132
},
{
"epoch": 0.1625916870415648,
"grad_norm": 0.703743577003479,
"learning_rate": 1.9680460581932447e-06,
"loss": 0.09523461759090424,
"step": 133
},
{
"epoch": 0.16381418092909536,
"grad_norm": 0.6345956325531006,
"learning_rate": 1.9675627180671944e-06,
"loss": 0.060631148517131805,
"step": 134
},
{
"epoch": 0.16503667481662593,
"grad_norm": 0.6716246008872986,
"learning_rate": 1.9670758100455355e-06,
"loss": 0.04715805873274803,
"step": 135
},
{
"epoch": 0.16625916870415647,
"grad_norm": 0.9317910671234131,
"learning_rate": 1.9665853359237447e-06,
"loss": 0.12580376863479614,
"step": 136
},
{
"epoch": 0.16748166259168704,
"grad_norm": 0.7225188612937927,
"learning_rate": 1.9660912975104498e-06,
"loss": 0.04970274493098259,
"step": 137
},
{
"epoch": 0.1687041564792176,
"grad_norm": 0.4728870093822479,
"learning_rate": 1.9655936966274213e-06,
"loss": 0.03771195188164711,
"step": 138
},
{
"epoch": 0.16992665036674817,
"grad_norm": 0.5576440095901489,
"learning_rate": 1.965092535109567e-06,
"loss": 0.05479685962200165,
"step": 139
},
{
"epoch": 0.17114914425427874,
"grad_norm": 0.6422237157821655,
"learning_rate": 1.9645878148049234e-06,
"loss": 0.0602768175303936,
"step": 140
},
{
"epoch": 0.17237163814180928,
"grad_norm": 0.8579259514808655,
"learning_rate": 1.964079537574652e-06,
"loss": 0.07493305206298828,
"step": 141
},
{
"epoch": 0.17359413202933985,
"grad_norm": 0.856762707233429,
"learning_rate": 1.9635677052930283e-06,
"loss": 0.08744668960571289,
"step": 142
},
{
"epoch": 0.17481662591687042,
"grad_norm": 0.8211008310317993,
"learning_rate": 1.9630523198474385e-06,
"loss": 0.06880424171686172,
"step": 143
},
{
"epoch": 0.17603911980440098,
"grad_norm": 0.5641351938247681,
"learning_rate": 1.962533383138371e-06,
"loss": 0.03602682054042816,
"step": 144
},
{
"epoch": 0.17726161369193155,
"grad_norm": 0.5784614086151123,
"learning_rate": 1.9620108970794085e-06,
"loss": 0.04793035238981247,
"step": 145
},
{
"epoch": 0.1784841075794621,
"grad_norm": 0.575289785861969,
"learning_rate": 1.961484863597224e-06,
"loss": 0.05398004874587059,
"step": 146
},
{
"epoch": 0.17970660146699266,
"grad_norm": 0.6905957460403442,
"learning_rate": 1.9609552846315693e-06,
"loss": 0.0722339078783989,
"step": 147
},
{
"epoch": 0.18092909535452323,
"grad_norm": 1.2074052095413208,
"learning_rate": 1.960422162135271e-06,
"loss": 0.20787547528743744,
"step": 148
},
{
"epoch": 0.1821515892420538,
"grad_norm": 0.8738341927528381,
"learning_rate": 1.959885498074224e-06,
"loss": 0.10185503214597702,
"step": 149
},
{
"epoch": 0.18337408312958436,
"grad_norm": 0.8463124632835388,
"learning_rate": 1.9593452944273804e-06,
"loss": 0.09371132403612137,
"step": 150
},
{
"epoch": 0.1845965770171149,
"grad_norm": 0.6591911315917969,
"learning_rate": 1.9588015531867464e-06,
"loss": 0.06103268265724182,
"step": 151
},
{
"epoch": 0.18581907090464547,
"grad_norm": 3.438685178756714,
"learning_rate": 1.9582542763573718e-06,
"loss": 0.22050483524799347,
"step": 152
},
{
"epoch": 0.18704156479217604,
"grad_norm": 6.808855056762695,
"learning_rate": 1.957703465957345e-06,
"loss": 0.2464558184146881,
"step": 153
},
{
"epoch": 0.1882640586797066,
"grad_norm": 4.114968776702881,
"learning_rate": 1.9571491240177832e-06,
"loss": 0.19404976069927216,
"step": 154
},
{
"epoch": 0.18948655256723718,
"grad_norm": 0.607691764831543,
"learning_rate": 1.9565912525828278e-06,
"loss": 0.061037201434373856,
"step": 155
},
{
"epoch": 0.19070904645476772,
"grad_norm": 0.7213531136512756,
"learning_rate": 1.9560298537096342e-06,
"loss": 0.07238991558551788,
"step": 156
},
{
"epoch": 0.19193154034229828,
"grad_norm": 0.6105731725692749,
"learning_rate": 1.955464929468365e-06,
"loss": 0.04562131315469742,
"step": 157
},
{
"epoch": 0.19315403422982885,
"grad_norm": 0.6376246213912964,
"learning_rate": 1.954896481942183e-06,
"loss": 0.06369476020336151,
"step": 158
},
{
"epoch": 0.19437652811735942,
"grad_norm": 3.05570125579834,
"learning_rate": 1.954324513227244e-06,
"loss": 0.1737847924232483,
"step": 159
},
{
"epoch": 0.19559902200489,
"grad_norm": 0.5501523017883301,
"learning_rate": 1.9537490254326863e-06,
"loss": 0.03715767338871956,
"step": 160
},
{
"epoch": 0.19682151589242053,
"grad_norm": 0.7043015956878662,
"learning_rate": 1.9531700206806273e-06,
"loss": 0.05609395354986191,
"step": 161
},
{
"epoch": 0.1980440097799511,
"grad_norm": 0.6382395029067993,
"learning_rate": 1.9525875011061506e-06,
"loss": 0.07138203084468842,
"step": 162
},
{
"epoch": 0.19926650366748166,
"grad_norm": 4.981842517852783,
"learning_rate": 1.9520014688573027e-06,
"loss": 0.28317272663116455,
"step": 163
},
{
"epoch": 0.20048899755501223,
"grad_norm": 0.5976245403289795,
"learning_rate": 1.951411926095082e-06,
"loss": 0.060227565467357635,
"step": 164
},
{
"epoch": 0.2017114914425428,
"grad_norm": 0.6383568048477173,
"learning_rate": 1.950818874993433e-06,
"loss": 0.06333554536104202,
"step": 165
},
{
"epoch": 0.20293398533007334,
"grad_norm": 0.5722646117210388,
"learning_rate": 1.950222317739236e-06,
"loss": 0.06169955059885979,
"step": 166
},
{
"epoch": 0.2041564792176039,
"grad_norm": 0.7577990293502808,
"learning_rate": 1.949622256532301e-06,
"loss": 0.06062877178192139,
"step": 167
},
{
"epoch": 0.20537897310513448,
"grad_norm": 0.5449824929237366,
"learning_rate": 1.949018693585359e-06,
"loss": 0.04920036718249321,
"step": 168
},
{
"epoch": 0.20660146699266504,
"grad_norm": 0.6604976058006287,
"learning_rate": 1.948411631124053e-06,
"loss": 0.07618685811758041,
"step": 169
},
{
"epoch": 0.2078239608801956,
"grad_norm": 0.9829358458518982,
"learning_rate": 1.9478010713869314e-06,
"loss": 0.11907469481229782,
"step": 170
},
{
"epoch": 0.20904645476772615,
"grad_norm": 1.0875358581542969,
"learning_rate": 1.9471870166254378e-06,
"loss": 0.07407742738723755,
"step": 171
},
{
"epoch": 0.21026894865525672,
"grad_norm": 0.6379914283752441,
"learning_rate": 1.9465694691039045e-06,
"loss": 0.07355710864067078,
"step": 172
},
{
"epoch": 0.2114914425427873,
"grad_norm": 0.981543242931366,
"learning_rate": 1.9459484310995428e-06,
"loss": 0.19537585973739624,
"step": 173
},
{
"epoch": 0.21271393643031786,
"grad_norm": 0.6272218823432922,
"learning_rate": 1.9453239049024362e-06,
"loss": 0.064679816365242,
"step": 174
},
{
"epoch": 0.2139364303178484,
"grad_norm": 1.4015697240829468,
"learning_rate": 1.94469589281553e-06,
"loss": 0.15909023582935333,
"step": 175
},
{
"epoch": 0.21515892420537897,
"grad_norm": 0.6106505393981934,
"learning_rate": 1.944064397154624e-06,
"loss": 0.05420103669166565,
"step": 176
},
{
"epoch": 0.21638141809290953,
"grad_norm": 0.6191409230232239,
"learning_rate": 1.943429420248363e-06,
"loss": 0.0645420178771019,
"step": 177
},
{
"epoch": 0.2176039119804401,
"grad_norm": 3.891395092010498,
"learning_rate": 1.942790964438231e-06,
"loss": 0.1691259890794754,
"step": 178
},
{
"epoch": 0.21882640586797067,
"grad_norm": 0.6322800517082214,
"learning_rate": 1.942149032078538e-06,
"loss": 0.03151990473270416,
"step": 179
},
{
"epoch": 0.2200488997555012,
"grad_norm": 0.7665536403656006,
"learning_rate": 1.941503625536416e-06,
"loss": 0.11028989404439926,
"step": 180
},
{
"epoch": 0.22127139364303178,
"grad_norm": 0.8057969212532043,
"learning_rate": 1.940854747191806e-06,
"loss": 0.10171699523925781,
"step": 181
},
{
"epoch": 0.22249388753056235,
"grad_norm": 0.5772876143455505,
"learning_rate": 1.940202399437453e-06,
"loss": 0.05573328584432602,
"step": 182
},
{
"epoch": 0.2237163814180929,
"grad_norm": 0.874727725982666,
"learning_rate": 1.939546584678894e-06,
"loss": 0.12126775830984116,
"step": 183
},
{
"epoch": 0.22493887530562348,
"grad_norm": 0.7666345238685608,
"learning_rate": 1.9388873053344524e-06,
"loss": 0.08826832473278046,
"step": 184
},
{
"epoch": 0.22616136919315402,
"grad_norm": 0.5564320087432861,
"learning_rate": 1.938224563835226e-06,
"loss": 0.035565149039030075,
"step": 185
},
{
"epoch": 0.2273838630806846,
"grad_norm": 0.8495666980743408,
"learning_rate": 1.9375583626250792e-06,
"loss": 0.1436379998922348,
"step": 186
},
{
"epoch": 0.22860635696821516,
"grad_norm": 0.5963088274002075,
"learning_rate": 1.936888704160635e-06,
"loss": 0.05425849184393883,
"step": 187
},
{
"epoch": 0.22982885085574573,
"grad_norm": 0.6368657946586609,
"learning_rate": 1.936215590911264e-06,
"loss": 0.05293988436460495,
"step": 188
},
{
"epoch": 0.2310513447432763,
"grad_norm": 0.8322920203208923,
"learning_rate": 1.935539025359077e-06,
"loss": 0.08429227769374847,
"step": 189
},
{
"epoch": 0.23227383863080683,
"grad_norm": 0.931011438369751,
"learning_rate": 1.9348590099989163e-06,
"loss": 0.07673133164644241,
"step": 190
},
{
"epoch": 0.2334963325183374,
"grad_norm": 0.5592967867851257,
"learning_rate": 1.934175547338343e-06,
"loss": 0.03433080390095711,
"step": 191
},
{
"epoch": 0.23471882640586797,
"grad_norm": 0.6808316707611084,
"learning_rate": 1.9334886398976315e-06,
"loss": 0.07346376031637192,
"step": 192
},
{
"epoch": 0.23594132029339854,
"grad_norm": 0.8805721402168274,
"learning_rate": 1.932798290209759e-06,
"loss": 0.0867384672164917,
"step": 193
},
{
"epoch": 0.2371638141809291,
"grad_norm": 0.7578855156898499,
"learning_rate": 1.932104500820396e-06,
"loss": 0.08246162533760071,
"step": 194
},
{
"epoch": 0.23838630806845965,
"grad_norm": 0.6729488372802734,
"learning_rate": 1.931407274287896e-06,
"loss": 0.08233211189508438,
"step": 195
},
{
"epoch": 0.2396088019559902,
"grad_norm": 5.295697212219238,
"learning_rate": 1.9307066131832884e-06,
"loss": 0.04929104447364807,
"step": 196
},
{
"epoch": 0.24083129584352078,
"grad_norm": 0.749836802482605,
"learning_rate": 1.930002520090266e-06,
"loss": 0.0616285502910614,
"step": 197
},
{
"epoch": 0.24205378973105135,
"grad_norm": 0.7304412126541138,
"learning_rate": 1.929294997605179e-06,
"loss": 0.07980424165725708,
"step": 198
},
{
"epoch": 0.24327628361858192,
"grad_norm": 0.4725780785083771,
"learning_rate": 1.928584048337022e-06,
"loss": 0.05388215184211731,
"step": 199
},
{
"epoch": 0.24449877750611246,
"grad_norm": 0.6431605815887451,
"learning_rate": 1.9278696749074257e-06,
"loss": 0.07114686071872711,
"step": 200
},
{
"epoch": 0.24572127139364303,
"grad_norm": 0.5807996392250061,
"learning_rate": 1.9271518799506488e-06,
"loss": 0.05719640105962753,
"step": 201
},
{
"epoch": 0.2469437652811736,
"grad_norm": 1.2892720699310303,
"learning_rate": 1.9264306661135663e-06,
"loss": 0.06708410382270813,
"step": 202
},
{
"epoch": 0.24816625916870416,
"grad_norm": 0.8082153797149658,
"learning_rate": 1.9257060360556606e-06,
"loss": 0.0794389471411705,
"step": 203
},
{
"epoch": 0.24938875305623473,
"grad_norm": 0.5927546620368958,
"learning_rate": 1.9249779924490096e-06,
"loss": 0.05324559658765793,
"step": 204
},
{
"epoch": 0.2506112469437653,
"grad_norm": 0.5051952600479126,
"learning_rate": 1.924246537978282e-06,
"loss": 0.03782281279563904,
"step": 205
},
{
"epoch": 0.25183374083129584,
"grad_norm": 0.5779122114181519,
"learning_rate": 1.923511675340721e-06,
"loss": 0.04000939056277275,
"step": 206
},
{
"epoch": 0.2530562347188264,
"grad_norm": 0.7431759834289551,
"learning_rate": 1.922773407246139e-06,
"loss": 0.0958084836602211,
"step": 207
},
{
"epoch": 0.254278728606357,
"grad_norm": 0.6318093538284302,
"learning_rate": 1.922031736416906e-06,
"loss": 0.061642758548259735,
"step": 208
},
{
"epoch": 0.2555012224938875,
"grad_norm": 0.5983513593673706,
"learning_rate": 1.9212866655879395e-06,
"loss": 0.05627179890871048,
"step": 209
},
{
"epoch": 0.2567237163814181,
"grad_norm": 0.664305567741394,
"learning_rate": 1.9205381975066943e-06,
"loss": 0.09234637767076492,
"step": 210
},
{
"epoch": 0.25794621026894865,
"grad_norm": 0.43261024355888367,
"learning_rate": 1.919786334933152e-06,
"loss": 0.047480352222919464,
"step": 211
},
{
"epoch": 0.2591687041564792,
"grad_norm": 0.7945061326026917,
"learning_rate": 1.919031080639813e-06,
"loss": 0.039106953889131546,
"step": 212
},
{
"epoch": 0.2603911980440098,
"grad_norm": 0.669981837272644,
"learning_rate": 1.9182724374116836e-06,
"loss": 0.04652997478842735,
"step": 213
},
{
"epoch": 0.2616136919315403,
"grad_norm": 2.418747663497925,
"learning_rate": 1.917510408046267e-06,
"loss": 0.19242990016937256,
"step": 214
},
{
"epoch": 0.2628361858190709,
"grad_norm": 0.5297693610191345,
"learning_rate": 1.916744995353553e-06,
"loss": 0.033531904220581055,
"step": 215
},
{
"epoch": 0.26405867970660146,
"grad_norm": 0.4767766296863556,
"learning_rate": 1.9159762021560068e-06,
"loss": 0.038153547793626785,
"step": 216
},
{
"epoch": 0.265281173594132,
"grad_norm": 0.4411314129829407,
"learning_rate": 1.91520403128856e-06,
"loss": 0.025550004094839096,
"step": 217
},
{
"epoch": 0.2665036674816626,
"grad_norm": 0.5952708721160889,
"learning_rate": 1.9144284855985996e-06,
"loss": 0.04932040348649025,
"step": 218
},
{
"epoch": 0.26772616136919314,
"grad_norm": 1.2439887523651123,
"learning_rate": 1.9136495679459563e-06,
"loss": 0.19842806458473206,
"step": 219
},
{
"epoch": 0.26894865525672373,
"grad_norm": 0.9705661535263062,
"learning_rate": 1.9128672812028955e-06,
"loss": 0.14921283721923828,
"step": 220
},
{
"epoch": 0.2701711491442543,
"grad_norm": 1.134079098701477,
"learning_rate": 1.9120816282541062e-06,
"loss": 0.05197975039482117,
"step": 221
},
{
"epoch": 0.2713936430317848,
"grad_norm": 0.8239823579788208,
"learning_rate": 1.91129261199669e-06,
"loss": 0.07877957075834274,
"step": 222
},
{
"epoch": 0.2726161369193154,
"grad_norm": 0.5756873488426208,
"learning_rate": 1.9105002353401514e-06,
"loss": 0.052862294018268585,
"step": 223
},
{
"epoch": 0.27383863080684595,
"grad_norm": 0.45038697123527527,
"learning_rate": 1.9097045012063853e-06,
"loss": 0.035664428025484085,
"step": 224
},
{
"epoch": 0.27506112469437655,
"grad_norm": 0.36812978982925415,
"learning_rate": 1.908905412529669e-06,
"loss": 0.04040788114070892,
"step": 225
},
{
"epoch": 0.2762836185819071,
"grad_norm": 0.5129841566085815,
"learning_rate": 1.9081029722566474e-06,
"loss": 0.056159667670726776,
"step": 226
},
{
"epoch": 0.2775061124694376,
"grad_norm": 3.171241283416748,
"learning_rate": 1.9072971833463268e-06,
"loss": 0.31288641691207886,
"step": 227
},
{
"epoch": 0.2787286063569682,
"grad_norm": 0.869947612285614,
"learning_rate": 1.9064880487700598e-06,
"loss": 0.07328721880912781,
"step": 228
},
{
"epoch": 0.27995110024449876,
"grad_norm": 0.6015138626098633,
"learning_rate": 1.9056755715115372e-06,
"loss": 0.07411491125822067,
"step": 229
},
{
"epoch": 0.28117359413202936,
"grad_norm": 0.5319263935089111,
"learning_rate": 1.9048597545667757e-06,
"loss": 0.05194948986172676,
"step": 230
},
{
"epoch": 0.2823960880195599,
"grad_norm": 0.5535708069801331,
"learning_rate": 1.904040600944107e-06,
"loss": 0.03731749579310417,
"step": 231
},
{
"epoch": 0.28361858190709044,
"grad_norm": 0.7307035326957703,
"learning_rate": 1.903218113664167e-06,
"loss": 0.1117473915219307,
"step": 232
},
{
"epoch": 0.28484107579462103,
"grad_norm": 0.6607651710510254,
"learning_rate": 1.9023922957598844e-06,
"loss": 0.08599244803190231,
"step": 233
},
{
"epoch": 0.2860635696821516,
"grad_norm": 0.7340969443321228,
"learning_rate": 1.9015631502764697e-06,
"loss": 0.09062845259904861,
"step": 234
},
{
"epoch": 0.28728606356968217,
"grad_norm": 0.39672523736953735,
"learning_rate": 1.9007306802714037e-06,
"loss": 0.031265828758478165,
"step": 235
},
{
"epoch": 0.2885085574572127,
"grad_norm": 0.619536817073822,
"learning_rate": 1.8998948888144265e-06,
"loss": 0.05924880504608154,
"step": 236
},
{
"epoch": 0.28973105134474325,
"grad_norm": 0.8397347927093506,
"learning_rate": 1.8990557789875263e-06,
"loss": 0.08695176243782043,
"step": 237
},
{
"epoch": 0.29095354523227385,
"grad_norm": 0.7448678612709045,
"learning_rate": 1.8982133538849274e-06,
"loss": 0.09118647873401642,
"step": 238
},
{
"epoch": 0.2921760391198044,
"grad_norm": 0.4982061982154846,
"learning_rate": 1.8973676166130791e-06,
"loss": 0.040331289172172546,
"step": 239
},
{
"epoch": 0.293398533007335,
"grad_norm": 0.7012085318565369,
"learning_rate": 1.8965185702906454e-06,
"loss": 0.07881244271993637,
"step": 240
},
{
"epoch": 0.2946210268948655,
"grad_norm": 0.4688606858253479,
"learning_rate": 1.8956662180484911e-06,
"loss": 0.030011724680662155,
"step": 241
},
{
"epoch": 0.29584352078239606,
"grad_norm": 0.5423737168312073,
"learning_rate": 1.8948105630296726e-06,
"loss": 0.06471644341945648,
"step": 242
},
{
"epoch": 0.29706601466992666,
"grad_norm": 1.0173776149749756,
"learning_rate": 1.8939516083894245e-06,
"loss": 0.11215247958898544,
"step": 243
},
{
"epoch": 0.2982885085574572,
"grad_norm": 0.606266438961029,
"learning_rate": 1.8930893572951497e-06,
"loss": 0.05849171802401543,
"step": 244
},
{
"epoch": 0.2995110024449878,
"grad_norm": 0.4643746316432953,
"learning_rate": 1.8922238129264057e-06,
"loss": 0.037738438695669174,
"step": 245
},
{
"epoch": 0.30073349633251834,
"grad_norm": 0.5321090221405029,
"learning_rate": 1.891354978474895e-06,
"loss": 0.03803202509880066,
"step": 246
},
{
"epoch": 0.3019559902200489,
"grad_norm": 0.7212055325508118,
"learning_rate": 1.8904828571444523e-06,
"loss": 0.06536910682916641,
"step": 247
},
{
"epoch": 0.30317848410757947,
"grad_norm": 0.6811750531196594,
"learning_rate": 1.8896074521510315e-06,
"loss": 0.07358380407094955,
"step": 248
},
{
"epoch": 0.30440097799511,
"grad_norm": 0.8438161015510559,
"learning_rate": 1.8887287667226963e-06,
"loss": 0.08441587537527084,
"step": 249
},
{
"epoch": 0.3056234718826406,
"grad_norm": 0.5656726360321045,
"learning_rate": 1.8878468040996065e-06,
"loss": 0.040838778018951416,
"step": 250
},
{
"epoch": 0.30684596577017115,
"grad_norm": 0.6735907793045044,
"learning_rate": 1.8869615675340065e-06,
"loss": 0.08724953979253769,
"step": 251
},
{
"epoch": 0.3080684596577017,
"grad_norm": 3.3891584873199463,
"learning_rate": 1.8860730602902137e-06,
"loss": 0.2965168356895447,
"step": 252
},
{
"epoch": 0.3092909535452323,
"grad_norm": 2.7479710578918457,
"learning_rate": 1.8851812856446058e-06,
"loss": 0.25384920835494995,
"step": 253
},
{
"epoch": 0.3105134474327628,
"grad_norm": 0.5872605443000793,
"learning_rate": 1.8842862468856096e-06,
"loss": 0.05429583415389061,
"step": 254
},
{
"epoch": 0.3117359413202934,
"grad_norm": 0.645785391330719,
"learning_rate": 1.8833879473136875e-06,
"loss": 0.08078612387180328,
"step": 255
},
{
"epoch": 0.31295843520782396,
"grad_norm": 0.563945472240448,
"learning_rate": 1.8824863902413273e-06,
"loss": 0.029984142631292343,
"step": 256
},
{
"epoch": 0.3141809290953545,
"grad_norm": 0.8126795887947083,
"learning_rate": 1.8815815789930273e-06,
"loss": 0.0959048792719841,
"step": 257
},
{
"epoch": 0.3154034229828851,
"grad_norm": 0.9309636354446411,
"learning_rate": 1.8806735169052875e-06,
"loss": 0.09874016791582108,
"step": 258
},
{
"epoch": 0.31662591687041564,
"grad_norm": 0.5479769706726074,
"learning_rate": 1.8797622073265943e-06,
"loss": 0.05605495721101761,
"step": 259
},
{
"epoch": 0.31784841075794623,
"grad_norm": 2.1465067863464355,
"learning_rate": 1.878847653617409e-06,
"loss": 0.17403331398963928,
"step": 260
},
{
"epoch": 0.31907090464547677,
"grad_norm": 0.5110053420066833,
"learning_rate": 1.8779298591501561e-06,
"loss": 0.04462679848074913,
"step": 261
},
{
"epoch": 0.3202933985330073,
"grad_norm": 0.9735310673713684,
"learning_rate": 1.8770088273092106e-06,
"loss": 0.11067380756139755,
"step": 262
},
{
"epoch": 0.3215158924205379,
"grad_norm": 0.6004728674888611,
"learning_rate": 1.876084561490885e-06,
"loss": 0.04293625056743622,
"step": 263
},
{
"epoch": 0.32273838630806845,
"grad_norm": 0.8620474934577942,
"learning_rate": 1.875157065103417e-06,
"loss": 0.06757684797048569,
"step": 264
},
{
"epoch": 0.32396088019559904,
"grad_norm": 0.5152825117111206,
"learning_rate": 1.8742263415669578e-06,
"loss": 0.02904273197054863,
"step": 265
},
{
"epoch": 0.3251833740831296,
"grad_norm": 0.7195943593978882,
"learning_rate": 1.8732923943135577e-06,
"loss": 0.06668753921985626,
"step": 266
},
{
"epoch": 0.3264058679706601,
"grad_norm": 0.8348450064659119,
"learning_rate": 1.8723552267871554e-06,
"loss": 0.07247181981801987,
"step": 267
},
{
"epoch": 0.3276283618581907,
"grad_norm": 4.533979415893555,
"learning_rate": 1.8714148424435637e-06,
"loss": 0.2647816240787506,
"step": 268
},
{
"epoch": 0.32885085574572126,
"grad_norm": 0.5464447140693665,
"learning_rate": 1.8704712447504579e-06,
"loss": 0.05539151653647423,
"step": 269
},
{
"epoch": 0.33007334963325186,
"grad_norm": 2.2506468296051025,
"learning_rate": 1.8695244371873619e-06,
"loss": 0.29975172877311707,
"step": 270
},
{
"epoch": 0.3312958435207824,
"grad_norm": 0.6142186522483826,
"learning_rate": 1.8685744232456373e-06,
"loss": 0.07224956899881363,
"step": 271
},
{
"epoch": 0.33251833740831294,
"grad_norm": 0.7161267399787903,
"learning_rate": 1.867621206428468e-06,
"loss": 0.09649151563644409,
"step": 272
},
{
"epoch": 0.33374083129584353,
"grad_norm": 0.7745113372802734,
"learning_rate": 1.8666647902508493e-06,
"loss": 0.06470485776662827,
"step": 273
},
{
"epoch": 0.33496332518337407,
"grad_norm": 0.8199692964553833,
"learning_rate": 1.865705178239573e-06,
"loss": 0.10250729322433472,
"step": 274
},
{
"epoch": 0.33618581907090467,
"grad_norm": 0.46994876861572266,
"learning_rate": 1.8647423739332173e-06,
"loss": 0.042732786387205124,
"step": 275
},
{
"epoch": 0.3374083129584352,
"grad_norm": 0.7389618158340454,
"learning_rate": 1.8637763808821307e-06,
"loss": 0.11835065484046936,
"step": 276
},
{
"epoch": 0.33863080684596575,
"grad_norm": 0.5606495141983032,
"learning_rate": 1.8628072026484211e-06,
"loss": 0.0480082705616951,
"step": 277
},
{
"epoch": 0.33985330073349634,
"grad_norm": 0.8801579475402832,
"learning_rate": 1.861834842805941e-06,
"loss": 0.08741495758295059,
"step": 278
},
{
"epoch": 0.3410757946210269,
"grad_norm": 0.6932021975517273,
"learning_rate": 1.8608593049402752e-06,
"loss": 0.08666648715734482,
"step": 279
},
{
"epoch": 0.3422982885085575,
"grad_norm": 0.6842429041862488,
"learning_rate": 1.859880592648728e-06,
"loss": 0.10341957956552505,
"step": 280
},
{
"epoch": 0.343520782396088,
"grad_norm": 0.8525963425636292,
"learning_rate": 1.858898709540309e-06,
"loss": 0.11454153805971146,
"step": 281
},
{
"epoch": 0.34474327628361856,
"grad_norm": 0.4129064083099365,
"learning_rate": 1.857913659235721e-06,
"loss": 0.03537704423069954,
"step": 282
},
{
"epoch": 0.34596577017114916,
"grad_norm": 0.8998320698738098,
"learning_rate": 1.8569254453673443e-06,
"loss": 0.1781051754951477,
"step": 283
},
{
"epoch": 0.3471882640586797,
"grad_norm": 0.7518042325973511,
"learning_rate": 1.8559340715792265e-06,
"loss": 0.11704909056425095,
"step": 284
},
{
"epoch": 0.3484107579462103,
"grad_norm": 0.7281680703163147,
"learning_rate": 1.8549395415270662e-06,
"loss": 0.05017225816845894,
"step": 285
},
{
"epoch": 0.34963325183374083,
"grad_norm": 0.6383973956108093,
"learning_rate": 1.8539418588782016e-06,
"loss": 0.07281845062971115,
"step": 286
},
{
"epoch": 0.3508557457212714,
"grad_norm": 0.4377782344818115,
"learning_rate": 1.852941027311596e-06,
"loss": 0.036647818982601166,
"step": 287
},
{
"epoch": 0.35207823960880197,
"grad_norm": 0.4793190658092499,
"learning_rate": 1.8519370505178238e-06,
"loss": 0.05104957893490791,
"step": 288
},
{
"epoch": 0.3533007334963325,
"grad_norm": 0.9575946927070618,
"learning_rate": 1.850929932199058e-06,
"loss": 0.11360795050859451,
"step": 289
},
{
"epoch": 0.3545232273838631,
"grad_norm": 2.2279927730560303,
"learning_rate": 1.8499196760690557e-06,
"loss": 0.14458535611629486,
"step": 290
},
{
"epoch": 0.35574572127139364,
"grad_norm": 0.8328719735145569,
"learning_rate": 1.8489062858531447e-06,
"loss": 0.08702249079942703,
"step": 291
},
{
"epoch": 0.3569682151589242,
"grad_norm": 0.7120832204818726,
"learning_rate": 1.8478897652882103e-06,
"loss": 0.06616386771202087,
"step": 292
},
{
"epoch": 0.3581907090464548,
"grad_norm": 0.8384228944778442,
"learning_rate": 1.8468701181226803e-06,
"loss": 0.11367875337600708,
"step": 293
},
{
"epoch": 0.3594132029339853,
"grad_norm": 0.6217339038848877,
"learning_rate": 1.845847348116512e-06,
"loss": 0.05224423110485077,
"step": 294
},
{
"epoch": 0.3606356968215159,
"grad_norm": 0.505452036857605,
"learning_rate": 1.8448214590411788e-06,
"loss": 0.04598171263933182,
"step": 295
},
{
"epoch": 0.36185819070904646,
"grad_norm": 0.5649967193603516,
"learning_rate": 1.8437924546796548e-06,
"loss": 0.08680104464292526,
"step": 296
},
{
"epoch": 0.363080684596577,
"grad_norm": 0.44298869371414185,
"learning_rate": 1.8427603388264023e-06,
"loss": 0.02976345643401146,
"step": 297
},
{
"epoch": 0.3643031784841076,
"grad_norm": 0.6897602081298828,
"learning_rate": 1.8417251152873574e-06,
"loss": 0.06644374877214432,
"step": 298
},
{
"epoch": 0.36552567237163813,
"grad_norm": 1.9302960634231567,
"learning_rate": 1.8406867878799152e-06,
"loss": 0.11358832567930222,
"step": 299
},
{
"epoch": 0.36674816625916873,
"grad_norm": 0.816354513168335,
"learning_rate": 1.839645360432917e-06,
"loss": 0.06752585619688034,
"step": 300
},
{
"epoch": 0.36797066014669927,
"grad_norm": 0.5234889388084412,
"learning_rate": 1.8386008367866348e-06,
"loss": 0.028082475066184998,
"step": 301
},
{
"epoch": 0.3691931540342298,
"grad_norm": 0.32663536071777344,
"learning_rate": 1.8375532207927588e-06,
"loss": 0.020044537261128426,
"step": 302
},
{
"epoch": 0.3704156479217604,
"grad_norm": 0.8540005683898926,
"learning_rate": 1.8365025163143813e-06,
"loss": 0.12166626751422882,
"step": 303
},
{
"epoch": 0.37163814180929094,
"grad_norm": 2.3167834281921387,
"learning_rate": 1.835448727225984e-06,
"loss": 0.1404266059398651,
"step": 304
},
{
"epoch": 0.37286063569682154,
"grad_norm": 2.4426651000976562,
"learning_rate": 1.834391857413423e-06,
"loss": 0.18857823312282562,
"step": 305
},
{
"epoch": 0.3740831295843521,
"grad_norm": 0.8191661834716797,
"learning_rate": 1.8333319107739145e-06,
"loss": 0.10840162634849548,
"step": 306
},
{
"epoch": 0.3753056234718826,
"grad_norm": 0.7946656346321106,
"learning_rate": 1.8322688912160206e-06,
"loss": 0.0873989462852478,
"step": 307
},
{
"epoch": 0.3765281173594132,
"grad_norm": 0.6423612833023071,
"learning_rate": 1.8312028026596349e-06,
"loss": 0.058997947722673416,
"step": 308
},
{
"epoch": 0.37775061124694376,
"grad_norm": 0.61423259973526,
"learning_rate": 1.8301336490359678e-06,
"loss": 0.040607307106256485,
"step": 309
},
{
"epoch": 0.37897310513447435,
"grad_norm": 0.6794923543930054,
"learning_rate": 1.8290614342875324e-06,
"loss": 0.10372830927371979,
"step": 310
},
{
"epoch": 0.3801955990220049,
"grad_norm": 0.6804410219192505,
"learning_rate": 1.8279861623681289e-06,
"loss": 0.11820713430643082,
"step": 311
},
{
"epoch": 0.38141809290953543,
"grad_norm": 0.6116454601287842,
"learning_rate": 1.826907837242832e-06,
"loss": 0.04397882521152496,
"step": 312
},
{
"epoch": 0.38264058679706603,
"grad_norm": 2.8428475856781006,
"learning_rate": 1.8258264628879751e-06,
"loss": 0.20437739789485931,
"step": 313
},
{
"epoch": 0.38386308068459657,
"grad_norm": 0.8590072989463806,
"learning_rate": 1.8247420432911341e-06,
"loss": 0.08183608949184418,
"step": 314
},
{
"epoch": 0.38508557457212717,
"grad_norm": 0.6727264523506165,
"learning_rate": 1.8236545824511164e-06,
"loss": 0.048170410096645355,
"step": 315
},
{
"epoch": 0.3863080684596577,
"grad_norm": 0.5671209096908569,
"learning_rate": 1.822564084377942e-06,
"loss": 0.037106212228536606,
"step": 316
},
{
"epoch": 0.38753056234718825,
"grad_norm": 0.7941469550132751,
"learning_rate": 1.8214705530928321e-06,
"loss": 0.11037755757570267,
"step": 317
},
{
"epoch": 0.38875305623471884,
"grad_norm": 0.5881237387657166,
"learning_rate": 1.8203739926281925e-06,
"loss": 0.06202580779790878,
"step": 318
},
{
"epoch": 0.3899755501222494,
"grad_norm": 0.6200165152549744,
"learning_rate": 1.819274407027599e-06,
"loss": 0.05761420354247093,
"step": 319
},
{
"epoch": 0.39119804400978,
"grad_norm": 0.5484597086906433,
"learning_rate": 1.8181718003457824e-06,
"loss": 0.03761240467429161,
"step": 320
},
{
"epoch": 0.3924205378973105,
"grad_norm": 0.6078844666481018,
"learning_rate": 1.8170661766486147e-06,
"loss": 0.05567587912082672,
"step": 321
},
{
"epoch": 0.39364303178484106,
"grad_norm": 0.7024797201156616,
"learning_rate": 1.815957540013092e-06,
"loss": 0.07020790874958038,
"step": 322
},
{
"epoch": 0.39486552567237165,
"grad_norm": 0.6443117260932922,
"learning_rate": 1.814845894527321e-06,
"loss": 0.06636602431535721,
"step": 323
},
{
"epoch": 0.3960880195599022,
"grad_norm": 0.8141049742698669,
"learning_rate": 1.8137312442905047e-06,
"loss": 0.06973614543676376,
"step": 324
},
{
"epoch": 0.3973105134474328,
"grad_norm": 1.44343900680542,
"learning_rate": 1.8126135934129238e-06,
"loss": 0.2302822768688202,
"step": 325
},
{
"epoch": 0.39853300733496333,
"grad_norm": 0.6236892342567444,
"learning_rate": 1.8114929460159265e-06,
"loss": 0.06303084641695023,
"step": 326
},
{
"epoch": 0.39975550122249387,
"grad_norm": 0.8395601511001587,
"learning_rate": 1.8103693062319088e-06,
"loss": 0.12374132126569748,
"step": 327
},
{
"epoch": 0.40097799511002447,
"grad_norm": 0.6072763800621033,
"learning_rate": 1.8092426782043018e-06,
"loss": 0.04017474129796028,
"step": 328
},
{
"epoch": 0.402200488997555,
"grad_norm": 0.4804489016532898,
"learning_rate": 1.8081130660875555e-06,
"loss": 0.0305608082562685,
"step": 329
},
{
"epoch": 0.4034229828850856,
"grad_norm": 0.4571274220943451,
"learning_rate": 1.8069804740471248e-06,
"loss": 0.03416018560528755,
"step": 330
},
{
"epoch": 0.40464547677261614,
"grad_norm": 0.6312482357025146,
"learning_rate": 1.805844906259452e-06,
"loss": 0.07479752600193024,
"step": 331
},
{
"epoch": 0.4058679706601467,
"grad_norm": 0.6809691190719604,
"learning_rate": 1.8047063669119522e-06,
"loss": 0.09651534259319305,
"step": 332
},
{
"epoch": 0.4070904645476773,
"grad_norm": 0.5043792128562927,
"learning_rate": 1.8035648602029995e-06,
"loss": 0.0441741868853569,
"step": 333
},
{
"epoch": 0.4083129584352078,
"grad_norm": 0.4679751992225647,
"learning_rate": 1.802420390341909e-06,
"loss": 0.046906523406505585,
"step": 334
},
{
"epoch": 0.4095354523227384,
"grad_norm": 0.8985747694969177,
"learning_rate": 1.8012729615489236e-06,
"loss": 0.12618057429790497,
"step": 335
},
{
"epoch": 0.41075794621026895,
"grad_norm": 0.7412994503974915,
"learning_rate": 1.8001225780551957e-06,
"loss": 0.08232929557561874,
"step": 336
},
{
"epoch": 0.4119804400977995,
"grad_norm": 1.1913223266601562,
"learning_rate": 1.7989692441027742e-06,
"loss": 0.19295917451381683,
"step": 337
},
{
"epoch": 0.4132029339853301,
"grad_norm": 0.8626223802566528,
"learning_rate": 1.797812963944588e-06,
"loss": 0.13327421247959137,
"step": 338
},
{
"epoch": 0.41442542787286063,
"grad_norm": 0.5332434773445129,
"learning_rate": 1.79665374184443e-06,
"loss": 0.03888559341430664,
"step": 339
},
{
"epoch": 0.4156479217603912,
"grad_norm": 0.5158253908157349,
"learning_rate": 1.7954915820769404e-06,
"loss": 0.04391035437583923,
"step": 340
},
{
"epoch": 0.41687041564792177,
"grad_norm": 0.8502625226974487,
"learning_rate": 1.7943264889275941e-06,
"loss": 0.06075948104262352,
"step": 341
},
{
"epoch": 0.4180929095354523,
"grad_norm": 0.4998021423816681,
"learning_rate": 1.7931584666926819e-06,
"loss": 0.02565477415919304,
"step": 342
},
{
"epoch": 0.4193154034229829,
"grad_norm": 0.5099864602088928,
"learning_rate": 1.7919875196792947e-06,
"loss": 0.045667439699172974,
"step": 343
},
{
"epoch": 0.42053789731051344,
"grad_norm": 0.48695844411849976,
"learning_rate": 1.7908136522053096e-06,
"loss": 0.031179869547486305,
"step": 344
},
{
"epoch": 0.421760391198044,
"grad_norm": 0.6427958607673645,
"learning_rate": 1.7896368685993735e-06,
"loss": 0.057823121547698975,
"step": 345
},
{
"epoch": 0.4229828850855746,
"grad_norm": 0.9684734344482422,
"learning_rate": 1.7884571732008847e-06,
"loss": 0.09379223734140396,
"step": 346
},
{
"epoch": 0.4242053789731051,
"grad_norm": 0.6612726449966431,
"learning_rate": 1.7872745703599804e-06,
"loss": 0.056322596967220306,
"step": 347
},
{
"epoch": 0.4254278728606357,
"grad_norm": 0.48835551738739014,
"learning_rate": 1.7860890644375188e-06,
"loss": 0.029711861163377762,
"step": 348
},
{
"epoch": 0.42665036674816625,
"grad_norm": 0.4879574179649353,
"learning_rate": 1.7849006598050625e-06,
"loss": 0.03095545992255211,
"step": 349
},
{
"epoch": 0.4278728606356968,
"grad_norm": 2.610914945602417,
"learning_rate": 1.7837093608448633e-06,
"loss": 0.13352835178375244,
"step": 350
},
{
"epoch": 0.4290953545232274,
"grad_norm": 0.6475844383239746,
"learning_rate": 1.7825151719498464e-06,
"loss": 0.07442446053028107,
"step": 351
},
{
"epoch": 0.43031784841075793,
"grad_norm": 0.6558871269226074,
"learning_rate": 1.7813180975235929e-06,
"loss": 0.06934991478919983,
"step": 352
},
{
"epoch": 0.4315403422982885,
"grad_norm": 0.7163611650466919,
"learning_rate": 1.7801181419803254e-06,
"loss": 0.09411974251270294,
"step": 353
},
{
"epoch": 0.43276283618581907,
"grad_norm": 0.6060804128646851,
"learning_rate": 1.7789153097448896e-06,
"loss": 0.060565024614334106,
"step": 354
},
{
"epoch": 0.4339853300733496,
"grad_norm": 0.8162394762039185,
"learning_rate": 1.7777096052527395e-06,
"loss": 0.09708255529403687,
"step": 355
},
{
"epoch": 0.4352078239608802,
"grad_norm": 0.609510064125061,
"learning_rate": 1.7765010329499205e-06,
"loss": 0.05245963856577873,
"step": 356
},
{
"epoch": 0.43643031784841074,
"grad_norm": 0.5424594283103943,
"learning_rate": 1.7752895972930535e-06,
"loss": 0.04589700698852539,
"step": 357
},
{
"epoch": 0.43765281173594134,
"grad_norm": 0.5115808248519897,
"learning_rate": 1.7740753027493165e-06,
"loss": 0.03695133328437805,
"step": 358
},
{
"epoch": 0.4388753056234719,
"grad_norm": 0.7606438994407654,
"learning_rate": 1.7728581537964318e-06,
"loss": 0.053554944694042206,
"step": 359
},
{
"epoch": 0.4400977995110024,
"grad_norm": 0.7404968738555908,
"learning_rate": 1.771638154922646e-06,
"loss": 0.08349598944187164,
"step": 360
},
{
"epoch": 0.441320293398533,
"grad_norm": 0.7266729474067688,
"learning_rate": 1.7704153106267149e-06,
"loss": 0.07703076303005219,
"step": 361
},
{
"epoch": 0.44254278728606355,
"grad_norm": 0.5275489687919617,
"learning_rate": 1.7691896254178876e-06,
"loss": 0.04531081020832062,
"step": 362
},
{
"epoch": 0.44376528117359415,
"grad_norm": 0.622312068939209,
"learning_rate": 1.7679611038158879e-06,
"loss": 0.06540070474147797,
"step": 363
},
{
"epoch": 0.4449877750611247,
"grad_norm": 0.5625421404838562,
"learning_rate": 1.7667297503508998e-06,
"loss": 0.060605164617300034,
"step": 364
},
{
"epoch": 0.44621026894865523,
"grad_norm": 0.6575480699539185,
"learning_rate": 1.7654955695635496e-06,
"loss": 0.0848669707775116,
"step": 365
},
{
"epoch": 0.4474327628361858,
"grad_norm": 0.6915276646614075,
"learning_rate": 1.7642585660048893e-06,
"loss": 0.06513918936252594,
"step": 366
},
{
"epoch": 0.44865525672371637,
"grad_norm": 0.6344141364097595,
"learning_rate": 1.7630187442363797e-06,
"loss": 0.08373470604419708,
"step": 367
},
{
"epoch": 0.44987775061124696,
"grad_norm": 0.6520289778709412,
"learning_rate": 1.761776108829874e-06,
"loss": 0.07537151873111725,
"step": 368
},
{
"epoch": 0.4511002444987775,
"grad_norm": 0.6743019223213196,
"learning_rate": 1.7605306643676006e-06,
"loss": 0.043923527002334595,
"step": 369
},
{
"epoch": 0.45232273838630804,
"grad_norm": 0.8348601460456848,
"learning_rate": 1.7592824154421462e-06,
"loss": 0.09336526691913605,
"step": 370
},
{
"epoch": 0.45354523227383864,
"grad_norm": 0.672042727470398,
"learning_rate": 1.7580313666564395e-06,
"loss": 0.06285228580236435,
"step": 371
},
{
"epoch": 0.4547677261613692,
"grad_norm": 0.8540533185005188,
"learning_rate": 1.756777522623733e-06,
"loss": 0.07249268889427185,
"step": 372
},
{
"epoch": 0.4559902200488998,
"grad_norm": 0.9703978300094604,
"learning_rate": 1.7555208879675874e-06,
"loss": 0.06379461288452148,
"step": 373
},
{
"epoch": 0.4572127139364303,
"grad_norm": 0.6504855751991272,
"learning_rate": 1.7542614673218532e-06,
"loss": 0.038014695048332214,
"step": 374
},
{
"epoch": 0.45843520782396086,
"grad_norm": 0.4223743677139282,
"learning_rate": 1.7529992653306547e-06,
"loss": 0.037842653691768646,
"step": 375
},
{
"epoch": 0.45965770171149145,
"grad_norm": 0.7640713453292847,
"learning_rate": 1.7517342866483723e-06,
"loss": 0.09737996757030487,
"step": 376
},
{
"epoch": 0.460880195599022,
"grad_norm": 0.5868320465087891,
"learning_rate": 1.7504665359396254e-06,
"loss": 0.05165889486670494,
"step": 377
},
{
"epoch": 0.4621026894865526,
"grad_norm": 1.9963880777359009,
"learning_rate": 1.7491960178792553e-06,
"loss": 0.13372284173965454,
"step": 378
},
{
"epoch": 0.4633251833740831,
"grad_norm": 0.5070524215698242,
"learning_rate": 1.747922737152308e-06,
"loss": 0.05226060748100281,
"step": 379
},
{
"epoch": 0.46454767726161367,
"grad_norm": 2.5499889850616455,
"learning_rate": 1.746646698454017e-06,
"loss": 0.1394466906785965,
"step": 380
},
{
"epoch": 0.46577017114914426,
"grad_norm": 0.4931853711605072,
"learning_rate": 1.7453679064897857e-06,
"loss": 0.024354252964258194,
"step": 381
},
{
"epoch": 0.4669926650366748,
"grad_norm": 0.8532901406288147,
"learning_rate": 1.7440863659751695e-06,
"loss": 0.12024171650409698,
"step": 382
},
{
"epoch": 0.4682151589242054,
"grad_norm": 0.41797932982444763,
"learning_rate": 1.7428020816358603e-06,
"loss": 0.03959495201706886,
"step": 383
},
{
"epoch": 0.46943765281173594,
"grad_norm": 0.6551591753959656,
"learning_rate": 1.7415150582076677e-06,
"loss": 0.09050138294696808,
"step": 384
},
{
"epoch": 0.4706601466992665,
"grad_norm": 0.4766867458820343,
"learning_rate": 1.7402253004365006e-06,
"loss": 0.03753257542848587,
"step": 385
},
{
"epoch": 0.4718826405867971,
"grad_norm": 0.5455687642097473,
"learning_rate": 1.7389328130783517e-06,
"loss": 0.0524420328438282,
"step": 386
},
{
"epoch": 0.4731051344743276,
"grad_norm": 0.5356265306472778,
"learning_rate": 1.7376376008992795e-06,
"loss": 0.046083614230155945,
"step": 387
},
{
"epoch": 0.4743276283618582,
"grad_norm": 0.8567264676094055,
"learning_rate": 1.736339668675389e-06,
"loss": 0.08800388127565384,
"step": 388
},
{
"epoch": 0.47555012224938875,
"grad_norm": 1.0687199831008911,
"learning_rate": 1.7350390211928166e-06,
"loss": 0.042738817632198334,
"step": 389
},
{
"epoch": 0.4767726161369193,
"grad_norm": 0.7236585021018982,
"learning_rate": 1.7337356632477107e-06,
"loss": 0.09903562813997269,
"step": 390
},
{
"epoch": 0.4779951100244499,
"grad_norm": 0.9953838586807251,
"learning_rate": 1.7324295996462144e-06,
"loss": 0.05916499346494675,
"step": 391
},
{
"epoch": 0.4792176039119804,
"grad_norm": 0.9180816411972046,
"learning_rate": 1.7311208352044485e-06,
"loss": 0.07151395827531815,
"step": 392
},
{
"epoch": 0.480440097799511,
"grad_norm": 0.629494309425354,
"learning_rate": 1.7298093747484922e-06,
"loss": 0.06647039204835892,
"step": 393
},
{
"epoch": 0.48166259168704156,
"grad_norm": 0.7529318928718567,
"learning_rate": 1.7284952231143674e-06,
"loss": 0.08990117907524109,
"step": 394
},
{
"epoch": 0.4828850855745721,
"grad_norm": 0.4904628396034241,
"learning_rate": 1.7271783851480192e-06,
"loss": 0.046744074672460556,
"step": 395
},
{
"epoch": 0.4841075794621027,
"grad_norm": 0.9402425289154053,
"learning_rate": 1.7258588657052985e-06,
"loss": 0.1316802203655243,
"step": 396
},
{
"epoch": 0.48533007334963324,
"grad_norm": 0.7497603297233582,
"learning_rate": 1.7245366696519446e-06,
"loss": 0.08589158952236176,
"step": 397
},
{
"epoch": 0.48655256723716384,
"grad_norm": 0.5577226281166077,
"learning_rate": 1.723211801863566e-06,
"loss": 0.049980975687503815,
"step": 398
},
{
"epoch": 0.4877750611246944,
"grad_norm": 0.6094166040420532,
"learning_rate": 1.721884267225624e-06,
"loss": 0.036646366119384766,
"step": 399
},
{
"epoch": 0.4889975550122249,
"grad_norm": 0.7852808833122253,
"learning_rate": 1.7205540706334135e-06,
"loss": 0.11063923686742783,
"step": 400
},
{
"epoch": 0.4902200488997555,
"grad_norm": 0.780125081539154,
"learning_rate": 1.7192212169920457e-06,
"loss": 0.07562121003866196,
"step": 401
},
{
"epoch": 0.49144254278728605,
"grad_norm": 2.034982919692993,
"learning_rate": 1.7178857112164295e-06,
"loss": 0.12301291525363922,
"step": 402
},
{
"epoch": 0.49266503667481665,
"grad_norm": 0.8010287880897522,
"learning_rate": 1.7165475582312537e-06,
"loss": 0.08541703224182129,
"step": 403
},
{
"epoch": 0.4938875305623472,
"grad_norm": 0.5602757930755615,
"learning_rate": 1.7152067629709683e-06,
"loss": 0.05827076733112335,
"step": 404
},
{
"epoch": 0.49511002444987773,
"grad_norm": 2.3077433109283447,
"learning_rate": 1.7138633303797674e-06,
"loss": 0.043225377798080444,
"step": 405
},
{
"epoch": 0.4963325183374083,
"grad_norm": 0.7154622673988342,
"learning_rate": 1.7125172654115699e-06,
"loss": 0.06698773801326752,
"step": 406
},
{
"epoch": 0.49755501222493886,
"grad_norm": 0.46365052461624146,
"learning_rate": 1.7111685730300013e-06,
"loss": 0.041519202291965485,
"step": 407
},
{
"epoch": 0.49877750611246946,
"grad_norm": 0.6418176293373108,
"learning_rate": 1.7098172582083763e-06,
"loss": 0.07674536108970642,
"step": 408
},
{
"epoch": 0.5,
"grad_norm": 0.4012064039707184,
"learning_rate": 1.7084633259296795e-06,
"loss": 0.03451547026634216,
"step": 409
},
{
"epoch": 0.5012224938875306,
"grad_norm": 0.646426260471344,
"learning_rate": 1.7071067811865474e-06,
"loss": 0.06606896221637726,
"step": 410
},
{
"epoch": 0.5024449877750611,
"grad_norm": 0.7879284620285034,
"learning_rate": 1.7057476289812502e-06,
"loss": 0.061796873807907104,
"step": 411
},
{
"epoch": 0.5036674816625917,
"grad_norm": 1.395188331604004,
"learning_rate": 1.7043858743256733e-06,
"loss": 0.16197964549064636,
"step": 412
},
{
"epoch": 0.5048899755501223,
"grad_norm": 1.3129609823226929,
"learning_rate": 1.7030215222412979e-06,
"loss": 0.09698643535375595,
"step": 413
},
{
"epoch": 0.5061124694376528,
"grad_norm": 0.5174209475517273,
"learning_rate": 1.7016545777591838e-06,
"loss": 0.046754635870456696,
"step": 414
},
{
"epoch": 0.5073349633251834,
"grad_norm": 0.7602027058601379,
"learning_rate": 1.7002850459199505e-06,
"loss": 0.10180725902318954,
"step": 415
},
{
"epoch": 0.508557457212714,
"grad_norm": 0.6113729476928711,
"learning_rate": 1.6989129317737576e-06,
"loss": 0.056210897862911224,
"step": 416
},
{
"epoch": 0.5097799511002445,
"grad_norm": 0.806771993637085,
"learning_rate": 1.6975382403802878e-06,
"loss": 0.07412853091955185,
"step": 417
},
{
"epoch": 0.511002444987775,
"grad_norm": 0.6962090730667114,
"learning_rate": 1.696160976808727e-06,
"loss": 0.10456901043653488,
"step": 418
},
{
"epoch": 0.5122249388753056,
"grad_norm": 1.0161210298538208,
"learning_rate": 1.6947811461377467e-06,
"loss": 0.07597371190786362,
"step": 419
},
{
"epoch": 0.5134474327628362,
"grad_norm": 0.7011575102806091,
"learning_rate": 1.693398753455483e-06,
"loss": 0.05035354942083359,
"step": 420
},
{
"epoch": 0.5146699266503667,
"grad_norm": 0.9028088450431824,
"learning_rate": 1.6920138038595212e-06,
"loss": 0.15767046809196472,
"step": 421
},
{
"epoch": 0.5158924205378973,
"grad_norm": 1.7831929922103882,
"learning_rate": 1.6906263024568744e-06,
"loss": 0.1559026837348938,
"step": 422
},
{
"epoch": 0.5171149144254279,
"grad_norm": 0.6349454522132874,
"learning_rate": 1.6892362543639653e-06,
"loss": 0.07751189917325974,
"step": 423
},
{
"epoch": 0.5183374083129584,
"grad_norm": 0.534195601940155,
"learning_rate": 1.6878436647066084e-06,
"loss": 0.04393431544303894,
"step": 424
},
{
"epoch": 0.519559902200489,
"grad_norm": 0.6865026354789734,
"learning_rate": 1.6864485386199891e-06,
"loss": 0.06122737377882004,
"step": 425
},
{
"epoch": 0.5207823960880196,
"grad_norm": 0.8534338474273682,
"learning_rate": 1.6850508812486467e-06,
"loss": 0.05949891731142998,
"step": 426
},
{
"epoch": 0.5220048899755502,
"grad_norm": 0.7156226634979248,
"learning_rate": 1.6836506977464548e-06,
"loss": 0.08566858619451523,
"step": 427
},
{
"epoch": 0.5232273838630807,
"grad_norm": 1.0050163269042969,
"learning_rate": 1.6822479932766017e-06,
"loss": 0.07646238803863525,
"step": 428
},
{
"epoch": 0.5244498777506112,
"grad_norm": 0.8601038455963135,
"learning_rate": 1.6808427730115712e-06,
"loss": 0.10100731253623962,
"step": 429
},
{
"epoch": 0.5256723716381418,
"grad_norm": 0.7451851963996887,
"learning_rate": 1.6794350421331257e-06,
"loss": 0.15804892778396606,
"step": 430
},
{
"epoch": 0.5268948655256723,
"grad_norm": 2.6624526977539062,
"learning_rate": 1.6780248058322837e-06,
"loss": 0.22299925982952118,
"step": 431
},
{
"epoch": 0.5281173594132029,
"grad_norm": 0.6954417824745178,
"learning_rate": 1.676612069309304e-06,
"loss": 0.06518401205539703,
"step": 432
},
{
"epoch": 0.5293398533007335,
"grad_norm": 0.4646545946598053,
"learning_rate": 1.6751968377736637e-06,
"loss": 0.03377952799201012,
"step": 433
},
{
"epoch": 0.530562347188264,
"grad_norm": 0.46790504455566406,
"learning_rate": 1.6737791164440416e-06,
"loss": 0.028043370693922043,
"step": 434
},
{
"epoch": 0.5317848410757946,
"grad_norm": 0.5663303732872009,
"learning_rate": 1.6723589105482966e-06,
"loss": 0.0559605211019516,
"step": 435
},
{
"epoch": 0.5330073349633252,
"grad_norm": 1.4141327142715454,
"learning_rate": 1.6709362253234499e-06,
"loss": 0.10892363637685776,
"step": 436
},
{
"epoch": 0.5342298288508558,
"grad_norm": 1.8901113271713257,
"learning_rate": 1.669511066015665e-06,
"loss": 0.14447565376758575,
"step": 437
},
{
"epoch": 0.5354523227383863,
"grad_norm": 0.7136497497558594,
"learning_rate": 1.6680834378802291e-06,
"loss": 0.07575707882642746,
"step": 438
},
{
"epoch": 0.5366748166259169,
"grad_norm": 0.6999083757400513,
"learning_rate": 1.6666533461815323e-06,
"loss": 0.08289598673582077,
"step": 439
},
{
"epoch": 0.5378973105134475,
"grad_norm": 0.879117488861084,
"learning_rate": 1.6652207961930505e-06,
"loss": 0.09258907288312912,
"step": 440
},
{
"epoch": 0.539119804400978,
"grad_norm": 0.6807286143302917,
"learning_rate": 1.663785793197323e-06,
"loss": 0.0701134204864502,
"step": 441
},
{
"epoch": 0.5403422982885085,
"grad_norm": 0.5867821574211121,
"learning_rate": 1.6623483424859358e-06,
"loss": 0.059491388499736786,
"step": 442
},
{
"epoch": 0.5415647921760391,
"grad_norm": 0.5489261746406555,
"learning_rate": 1.6609084493595e-06,
"loss": 0.07165656983852386,
"step": 443
},
{
"epoch": 0.5427872860635696,
"grad_norm": 1.3966374397277832,
"learning_rate": 1.659466119127634e-06,
"loss": 0.2164561003446579,
"step": 444
},
{
"epoch": 0.5440097799511002,
"grad_norm": 0.5128722190856934,
"learning_rate": 1.6580213571089425e-06,
"loss": 0.018365073949098587,
"step": 445
},
{
"epoch": 0.5452322738386308,
"grad_norm": 0.8345523476600647,
"learning_rate": 1.656574168630997e-06,
"loss": 0.14947578310966492,
"step": 446
},
{
"epoch": 0.5464547677261614,
"grad_norm": 0.5018194317817688,
"learning_rate": 1.6551245590303177e-06,
"loss": 0.04533746466040611,
"step": 447
},
{
"epoch": 0.5476772616136919,
"grad_norm": 0.42639148235321045,
"learning_rate": 1.6536725336523519e-06,
"loss": 0.02630656585097313,
"step": 448
},
{
"epoch": 0.5488997555012225,
"grad_norm": 1.7148997783660889,
"learning_rate": 1.6522180978514552e-06,
"loss": 0.15345633029937744,
"step": 449
},
{
"epoch": 0.5501222493887531,
"grad_norm": 0.9442937970161438,
"learning_rate": 1.6507612569908723e-06,
"loss": 0.11478523910045624,
"step": 450
},
{
"epoch": 0.5513447432762836,
"grad_norm": 0.6311274766921997,
"learning_rate": 1.6493020164427153e-06,
"loss": 0.07657258212566376,
"step": 451
},
{
"epoch": 0.5525672371638142,
"grad_norm": 0.6714059114456177,
"learning_rate": 1.6478403815879465e-06,
"loss": 0.07234999537467957,
"step": 452
},
{
"epoch": 0.5537897310513448,
"grad_norm": 0.46692174673080444,
"learning_rate": 1.646376357816356e-06,
"loss": 0.038375988602638245,
"step": 453
},
{
"epoch": 0.5550122249388753,
"grad_norm": 0.6291329860687256,
"learning_rate": 1.6449099505265445e-06,
"loss": 0.07796897739171982,
"step": 454
},
{
"epoch": 0.5562347188264058,
"grad_norm": 0.6769044995307922,
"learning_rate": 1.6434411651259006e-06,
"loss": 0.04393567144870758,
"step": 455
},
{
"epoch": 0.5574572127139364,
"grad_norm": 0.7879950404167175,
"learning_rate": 1.6419700070305828e-06,
"loss": 0.08492709696292877,
"step": 456
},
{
"epoch": 0.558679706601467,
"grad_norm": 1.1804296970367432,
"learning_rate": 1.6404964816654992e-06,
"loss": 0.1638970524072647,
"step": 457
},
{
"epoch": 0.5599022004889975,
"grad_norm": 0.5166059732437134,
"learning_rate": 1.6390205944642863e-06,
"loss": 0.05668896436691284,
"step": 458
},
{
"epoch": 0.5611246943765281,
"grad_norm": 0.6031139492988586,
"learning_rate": 1.6375423508692912e-06,
"loss": 0.05560121685266495,
"step": 459
},
{
"epoch": 0.5623471882640587,
"grad_norm": 3.1707417964935303,
"learning_rate": 1.6360617563315489e-06,
"loss": 0.16623622179031372,
"step": 460
},
{
"epoch": 0.5635696821515892,
"grad_norm": 1.5755645036697388,
"learning_rate": 1.6345788163107643e-06,
"loss": 0.03788503631949425,
"step": 461
},
{
"epoch": 0.5647921760391198,
"grad_norm": 0.5728036761283875,
"learning_rate": 1.6330935362752914e-06,
"loss": 0.028920333832502365,
"step": 462
},
{
"epoch": 0.5660146699266504,
"grad_norm": 0.5930987000465393,
"learning_rate": 1.6316059217021123e-06,
"loss": 0.05874459445476532,
"step": 463
},
{
"epoch": 0.5672371638141809,
"grad_norm": 0.9829890131950378,
"learning_rate": 1.6301159780768187e-06,
"loss": 0.1674955189228058,
"step": 464
},
{
"epoch": 0.5684596577017115,
"grad_norm": 0.6058177351951599,
"learning_rate": 1.62862371089359e-06,
"loss": 0.0640096515417099,
"step": 465
},
{
"epoch": 0.5696821515892421,
"grad_norm": 0.5851539969444275,
"learning_rate": 1.6271291256551734e-06,
"loss": 0.06367719918489456,
"step": 466
},
{
"epoch": 0.5709046454767727,
"grad_norm": 0.5370099544525146,
"learning_rate": 1.625632227872865e-06,
"loss": 0.029538314789533615,
"step": 467
},
{
"epoch": 0.5721271393643031,
"grad_norm": 0.6615535020828247,
"learning_rate": 1.6241330230664877e-06,
"loss": 0.03103392943739891,
"step": 468
},
{
"epoch": 0.5733496332518337,
"grad_norm": 0.6154413223266602,
"learning_rate": 1.622631516764372e-06,
"loss": 0.07144662737846375,
"step": 469
},
{
"epoch": 0.5745721271393643,
"grad_norm": 0.7187280654907227,
"learning_rate": 1.6211277145033347e-06,
"loss": 0.06100647896528244,
"step": 470
},
{
"epoch": 0.5757946210268948,
"grad_norm": 0.5423647165298462,
"learning_rate": 1.619621621828659e-06,
"loss": 0.058304429054260254,
"step": 471
},
{
"epoch": 0.5770171149144254,
"grad_norm": 0.8790974020957947,
"learning_rate": 1.6181132442940746e-06,
"loss": 0.08741569519042969,
"step": 472
},
{
"epoch": 0.578239608801956,
"grad_norm": 0.4192120134830475,
"learning_rate": 1.616602587461736e-06,
"loss": 0.03600487485527992,
"step": 473
},
{
"epoch": 0.5794621026894865,
"grad_norm": 1.0022236108779907,
"learning_rate": 1.615089656902203e-06,
"loss": 0.15466028451919556,
"step": 474
},
{
"epoch": 0.5806845965770171,
"grad_norm": 0.6095752120018005,
"learning_rate": 1.6135744581944199e-06,
"loss": 0.07956439256668091,
"step": 475
},
{
"epoch": 0.5819070904645477,
"grad_norm": 1.5193233489990234,
"learning_rate": 1.6120569969256942e-06,
"loss": 0.21943241357803345,
"step": 476
},
{
"epoch": 0.5831295843520783,
"grad_norm": 0.6240202188491821,
"learning_rate": 1.6105372786916774e-06,
"loss": 0.06339319050312042,
"step": 477
},
{
"epoch": 0.5843520782396088,
"grad_norm": 0.6270899176597595,
"learning_rate": 1.6090153090963425e-06,
"loss": 0.0648411437869072,
"step": 478
},
{
"epoch": 0.5855745721271394,
"grad_norm": 0.45846807956695557,
"learning_rate": 1.607491093751966e-06,
"loss": 0.02837253361940384,
"step": 479
},
{
"epoch": 0.58679706601467,
"grad_norm": 0.4432011544704437,
"learning_rate": 1.6059646382791046e-06,
"loss": 0.03572048991918564,
"step": 480
},
{
"epoch": 0.5880195599022005,
"grad_norm": 2.140742063522339,
"learning_rate": 1.604435948306575e-06,
"loss": 0.1341155767440796,
"step": 481
},
{
"epoch": 0.589242053789731,
"grad_norm": 0.6492090225219727,
"learning_rate": 1.6029050294714347e-06,
"loss": 0.0744350254535675,
"step": 482
},
{
"epoch": 0.5904645476772616,
"grad_norm": 0.7156934142112732,
"learning_rate": 1.6013718874189592e-06,
"loss": 0.12815822660923004,
"step": 483
},
{
"epoch": 0.5916870415647921,
"grad_norm": 0.6490713357925415,
"learning_rate": 1.5998365278026232e-06,
"loss": 0.08208261430263519,
"step": 484
},
{
"epoch": 0.5929095354523227,
"grad_norm": 0.7055416107177734,
"learning_rate": 1.5982989562840784e-06,
"loss": 0.048287637531757355,
"step": 485
},
{
"epoch": 0.5941320293398533,
"grad_norm": 0.9537354707717896,
"learning_rate": 1.5967591785331313e-06,
"loss": 0.11526753008365631,
"step": 486
},
{
"epoch": 0.5953545232273839,
"grad_norm": 0.798416018486023,
"learning_rate": 1.5952172002277266e-06,
"loss": 0.09059370309114456,
"step": 487
},
{
"epoch": 0.5965770171149144,
"grad_norm": 0.5278511643409729,
"learning_rate": 1.5936730270539217e-06,
"loss": 0.028639772906899452,
"step": 488
},
{
"epoch": 0.597799511002445,
"grad_norm": 0.7626193761825562,
"learning_rate": 1.592126664705868e-06,
"loss": 0.10248832404613495,
"step": 489
},
{
"epoch": 0.5990220048899756,
"grad_norm": 0.8720624446868896,
"learning_rate": 1.5905781188857895e-06,
"loss": 0.14895012974739075,
"step": 490
},
{
"epoch": 0.6002444987775061,
"grad_norm": 0.366016149520874,
"learning_rate": 1.589027395303962e-06,
"loss": 0.02492983266711235,
"step": 491
},
{
"epoch": 0.6014669926650367,
"grad_norm": 0.6702889800071716,
"learning_rate": 1.5874744996786915e-06,
"loss": 0.06345844268798828,
"step": 492
},
{
"epoch": 0.6026894865525673,
"grad_norm": 0.46186578273773193,
"learning_rate": 1.5859194377362938e-06,
"loss": 0.034542396664619446,
"step": 493
},
{
"epoch": 0.6039119804400978,
"grad_norm": 0.6209185123443604,
"learning_rate": 1.5843622152110729e-06,
"loss": 0.04155988246202469,
"step": 494
},
{
"epoch": 0.6051344743276283,
"grad_norm": 1.9965450763702393,
"learning_rate": 1.5828028378452998e-06,
"loss": 0.21958570182323456,
"step": 495
},
{
"epoch": 0.6063569682151589,
"grad_norm": 0.5680025815963745,
"learning_rate": 1.5812413113891916e-06,
"loss": 0.07041500508785248,
"step": 496
},
{
"epoch": 0.6075794621026895,
"grad_norm": 0.7814162373542786,
"learning_rate": 1.5796776416008896e-06,
"loss": 0.11981351673603058,
"step": 497
},
{
"epoch": 0.60880195599022,
"grad_norm": 0.752768874168396,
"learning_rate": 1.5781118342464401e-06,
"loss": 0.10103460401296616,
"step": 498
},
{
"epoch": 0.6100244498777506,
"grad_norm": 0.8338395357131958,
"learning_rate": 1.5765438950997703e-06,
"loss": 0.09526871144771576,
"step": 499
},
{
"epoch": 0.6112469437652812,
"grad_norm": 0.6249001622200012,
"learning_rate": 1.5749738299426687e-06,
"loss": 0.06810910254716873,
"step": 500
},
{
"epoch": 0.6124694376528117,
"grad_norm": 0.417657732963562,
"learning_rate": 1.573401644564764e-06,
"loss": 0.026094753295183182,
"step": 501
},
{
"epoch": 0.6136919315403423,
"grad_norm": 1.5352287292480469,
"learning_rate": 1.5718273447635032e-06,
"loss": 0.21469885110855103,
"step": 502
},
{
"epoch": 0.6149144254278729,
"grad_norm": 0.9674742221832275,
"learning_rate": 1.5702509363441295e-06,
"loss": 0.09993942081928253,
"step": 503
},
{
"epoch": 0.6161369193154034,
"grad_norm": 0.5614142417907715,
"learning_rate": 1.5686724251196622e-06,
"loss": 0.05615081638097763,
"step": 504
},
{
"epoch": 0.617359413202934,
"grad_norm": 0.9605313539505005,
"learning_rate": 1.5670918169108748e-06,
"loss": 0.0903533399105072,
"step": 505
},
{
"epoch": 0.6185819070904646,
"grad_norm": 0.5507572889328003,
"learning_rate": 1.5655091175462732e-06,
"loss": 0.03787444904446602,
"step": 506
},
{
"epoch": 0.6198044009779952,
"grad_norm": 0.6130024194717407,
"learning_rate": 1.5639243328620743e-06,
"loss": 0.05924457684159279,
"step": 507
},
{
"epoch": 0.6210268948655256,
"grad_norm": 0.7767773866653442,
"learning_rate": 1.5623374687021853e-06,
"loss": 0.09556356072425842,
"step": 508
},
{
"epoch": 0.6222493887530562,
"grad_norm": 0.77219158411026,
"learning_rate": 1.5607485309181812e-06,
"loss": 0.12650828063488007,
"step": 509
},
{
"epoch": 0.6234718826405868,
"grad_norm": 0.7459167838096619,
"learning_rate": 1.5591575253692827e-06,
"loss": 0.06710905581712723,
"step": 510
},
{
"epoch": 0.6246943765281173,
"grad_norm": 0.5183116793632507,
"learning_rate": 1.557564457922336e-06,
"loss": 0.05422871559858322,
"step": 511
},
{
"epoch": 0.6259168704156479,
"grad_norm": 0.5681304931640625,
"learning_rate": 1.5559693344517914e-06,
"loss": 0.0623016394674778,
"step": 512
},
{
"epoch": 0.6271393643031785,
"grad_norm": 0.5745004415512085,
"learning_rate": 1.5543721608396796e-06,
"loss": 0.07345229387283325,
"step": 513
},
{
"epoch": 0.628361858190709,
"grad_norm": 0.6944248080253601,
"learning_rate": 1.5527729429755915e-06,
"loss": 0.05411592125892639,
"step": 514
},
{
"epoch": 0.6295843520782396,
"grad_norm": 0.8710147142410278,
"learning_rate": 1.5511716867566567e-06,
"loss": 0.11725334823131561,
"step": 515
},
{
"epoch": 0.6308068459657702,
"grad_norm": 0.4799576699733734,
"learning_rate": 1.5495683980875204e-06,
"loss": 0.037959832698106766,
"step": 516
},
{
"epoch": 0.6320293398533008,
"grad_norm": 0.7103644609451294,
"learning_rate": 1.5479630828803234e-06,
"loss": 0.063499815762043,
"step": 517
},
{
"epoch": 0.6332518337408313,
"grad_norm": 0.5715715289115906,
"learning_rate": 1.5463557470546785e-06,
"loss": 0.06349049508571625,
"step": 518
},
{
"epoch": 0.6344743276283619,
"grad_norm": 0.9053394198417664,
"learning_rate": 1.544746396537651e-06,
"loss": 0.08265362679958344,
"step": 519
},
{
"epoch": 0.6356968215158925,
"grad_norm": 0.4537986218929291,
"learning_rate": 1.5431350372637333e-06,
"loss": 0.040675289928913116,
"step": 520
},
{
"epoch": 0.636919315403423,
"grad_norm": 0.9152933359146118,
"learning_rate": 1.5415216751748263e-06,
"loss": 0.08792457729578018,
"step": 521
},
{
"epoch": 0.6381418092909535,
"grad_norm": 0.512662410736084,
"learning_rate": 1.5399063162202168e-06,
"loss": 0.04165332019329071,
"step": 522
},
{
"epoch": 0.6393643031784841,
"grad_norm": 0.9734593629837036,
"learning_rate": 1.538288966356554e-06,
"loss": 0.10679678618907928,
"step": 523
},
{
"epoch": 0.6405867970660146,
"grad_norm": 0.8143560886383057,
"learning_rate": 1.5366696315478288e-06,
"loss": 0.06281153857707977,
"step": 524
},
{
"epoch": 0.6418092909535452,
"grad_norm": 0.9854868054389954,
"learning_rate": 1.5350483177653527e-06,
"loss": 0.11162212491035461,
"step": 525
},
{
"epoch": 0.6430317848410758,
"grad_norm": 2.0043885707855225,
"learning_rate": 1.533425030987733e-06,
"loss": 0.13091318309307098,
"step": 526
},
{
"epoch": 0.6442542787286064,
"grad_norm": 0.670173168182373,
"learning_rate": 1.5317997772008536e-06,
"loss": 0.06878118216991425,
"step": 527
},
{
"epoch": 0.6454767726161369,
"grad_norm": 0.8089818954467773,
"learning_rate": 1.5301725623978511e-06,
"loss": 0.06651976704597473,
"step": 528
},
{
"epoch": 0.6466992665036675,
"grad_norm": 1.8312727212905884,
"learning_rate": 1.5285433925790945e-06,
"loss": 0.18683244287967682,
"step": 529
},
{
"epoch": 0.6479217603911981,
"grad_norm": 0.29744306206703186,
"learning_rate": 1.5269122737521603e-06,
"loss": 0.01824149303138256,
"step": 530
},
{
"epoch": 0.6491442542787286,
"grad_norm": 0.9314236044883728,
"learning_rate": 1.5252792119318128e-06,
"loss": 0.10050943493843079,
"step": 531
},
{
"epoch": 0.6503667481662592,
"grad_norm": 0.6798034906387329,
"learning_rate": 1.5236442131399818e-06,
"loss": 0.08808783441781998,
"step": 532
},
{
"epoch": 0.6515892420537898,
"grad_norm": 0.5657051205635071,
"learning_rate": 1.5220072834057386e-06,
"loss": 0.075182244181633,
"step": 533
},
{
"epoch": 0.6528117359413202,
"grad_norm": 1.4526220560073853,
"learning_rate": 1.520368428765275e-06,
"loss": 0.14190104603767395,
"step": 534
},
{
"epoch": 0.6540342298288508,
"grad_norm": 1.529256820678711,
"learning_rate": 1.5187276552618816e-06,
"loss": 0.13207709789276123,
"step": 535
},
{
"epoch": 0.6552567237163814,
"grad_norm": 1.7788233757019043,
"learning_rate": 1.5170849689459244e-06,
"loss": 0.16790451109409332,
"step": 536
},
{
"epoch": 0.656479217603912,
"grad_norm": 0.8590500354766846,
"learning_rate": 1.5154403758748226e-06,
"loss": 0.13659724593162537,
"step": 537
},
{
"epoch": 0.6577017114914425,
"grad_norm": 0.4188726842403412,
"learning_rate": 1.513793882113027e-06,
"loss": 0.04456861689686775,
"step": 538
},
{
"epoch": 0.6589242053789731,
"grad_norm": 0.6384551525115967,
"learning_rate": 1.5121454937319975e-06,
"loss": 0.05087222903966904,
"step": 539
},
{
"epoch": 0.6601466992665037,
"grad_norm": 0.6052460074424744,
"learning_rate": 1.5104952168101796e-06,
"loss": 0.08869906514883041,
"step": 540
},
{
"epoch": 0.6613691931540342,
"grad_norm": 0.5958714485168457,
"learning_rate": 1.5088430574329834e-06,
"loss": 0.04052342474460602,
"step": 541
},
{
"epoch": 0.6625916870415648,
"grad_norm": 0.6830777525901794,
"learning_rate": 1.5071890216927603e-06,
"loss": 0.059434644877910614,
"step": 542
},
{
"epoch": 0.6638141809290954,
"grad_norm": 0.6288948655128479,
"learning_rate": 1.505533115688781e-06,
"loss": 0.08509357273578644,
"step": 543
},
{
"epoch": 0.6650366748166259,
"grad_norm": 0.5015953779220581,
"learning_rate": 1.5038753455272123e-06,
"loss": 0.03578415513038635,
"step": 544
},
{
"epoch": 0.6662591687041565,
"grad_norm": 0.8920609951019287,
"learning_rate": 1.5022157173210968e-06,
"loss": 0.1283036172389984,
"step": 545
},
{
"epoch": 0.6674816625916871,
"grad_norm": 1.1573212146759033,
"learning_rate": 1.5005542371903258e-06,
"loss": 0.1203259527683258,
"step": 546
},
{
"epoch": 0.6687041564792175,
"grad_norm": 0.747620165348053,
"learning_rate": 1.498890911261622e-06,
"loss": 0.06897878646850586,
"step": 547
},
{
"epoch": 0.6699266503667481,
"grad_norm": 0.7635433673858643,
"learning_rate": 1.4972257456685137e-06,
"loss": 0.10960714519023895,
"step": 548
},
{
"epoch": 0.6711491442542787,
"grad_norm": 0.6831878423690796,
"learning_rate": 1.4955587465513128e-06,
"loss": 0.07159101217985153,
"step": 549
},
{
"epoch": 0.6723716381418093,
"grad_norm": 0.561427652835846,
"learning_rate": 1.4938899200570924e-06,
"loss": 0.05123119056224823,
"step": 550
},
{
"epoch": 0.6735941320293398,
"grad_norm": 0.5385869741439819,
"learning_rate": 1.4922192723396644e-06,
"loss": 0.038459889590740204,
"step": 551
},
{
"epoch": 0.6748166259168704,
"grad_norm": 0.8276560306549072,
"learning_rate": 1.490546809559556e-06,
"loss": 0.09098070859909058,
"step": 552
},
{
"epoch": 0.676039119804401,
"grad_norm": 0.4349839985370636,
"learning_rate": 1.4888725378839876e-06,
"loss": 0.02071479707956314,
"step": 553
},
{
"epoch": 0.6772616136919315,
"grad_norm": 0.586810827255249,
"learning_rate": 1.4871964634868499e-06,
"loss": 0.0504005067050457,
"step": 554
},
{
"epoch": 0.6784841075794621,
"grad_norm": 0.7760816812515259,
"learning_rate": 1.4855185925486818e-06,
"loss": 0.07645124942064285,
"step": 555
},
{
"epoch": 0.6797066014669927,
"grad_norm": 0.7232956886291504,
"learning_rate": 1.4838389312566457e-06,
"loss": 0.10397888720035553,
"step": 556
},
{
"epoch": 0.6809290953545232,
"grad_norm": 0.5987189412117004,
"learning_rate": 1.4821574858045072e-06,
"loss": 0.05084611102938652,
"step": 557
},
{
"epoch": 0.6821515892420538,
"grad_norm": 0.7198056578636169,
"learning_rate": 1.4804742623926104e-06,
"loss": 0.05376545339822769,
"step": 558
},
{
"epoch": 0.6833740831295844,
"grad_norm": 0.7698935866355896,
"learning_rate": 1.4787892672278555e-06,
"loss": 0.10251826047897339,
"step": 559
},
{
"epoch": 0.684596577017115,
"grad_norm": 0.8758496046066284,
"learning_rate": 1.4771025065236762e-06,
"loss": 0.15007014572620392,
"step": 560
},
{
"epoch": 0.6858190709046454,
"grad_norm": 0.6506555676460266,
"learning_rate": 1.475413986500017e-06,
"loss": 0.08068202435970306,
"step": 561
},
{
"epoch": 0.687041564792176,
"grad_norm": 0.4701099693775177,
"learning_rate": 1.4737237133833094e-06,
"loss": 0.042859457433223724,
"step": 562
},
{
"epoch": 0.6882640586797066,
"grad_norm": 0.5884132981300354,
"learning_rate": 1.4720316934064493e-06,
"loss": 0.052832458168268204,
"step": 563
},
{
"epoch": 0.6894865525672371,
"grad_norm": 0.6812411546707153,
"learning_rate": 1.4703379328087753e-06,
"loss": 0.0579126700758934,
"step": 564
},
{
"epoch": 0.6907090464547677,
"grad_norm": 1.3013073205947876,
"learning_rate": 1.4686424378360432e-06,
"loss": 0.17248809337615967,
"step": 565
},
{
"epoch": 0.6919315403422983,
"grad_norm": 0.6550065279006958,
"learning_rate": 1.4669452147404048e-06,
"loss": 0.03864119201898575,
"step": 566
},
{
"epoch": 0.6931540342298288,
"grad_norm": 0.9103739261627197,
"learning_rate": 1.4652462697803848e-06,
"loss": 0.10695357620716095,
"step": 567
},
{
"epoch": 0.6943765281173594,
"grad_norm": 0.32174891233444214,
"learning_rate": 1.4635456092208569e-06,
"loss": 0.021962780505418777,
"step": 568
},
{
"epoch": 0.69559902200489,
"grad_norm": 0.8829606175422668,
"learning_rate": 1.461843239333021e-06,
"loss": 0.09130948781967163,
"step": 569
},
{
"epoch": 0.6968215158924206,
"grad_norm": 0.4510987401008606,
"learning_rate": 1.46013916639438e-06,
"loss": 0.03176015987992287,
"step": 570
},
{
"epoch": 0.6980440097799511,
"grad_norm": 0.5968968868255615,
"learning_rate": 1.4584333966887177e-06,
"loss": 0.06604845821857452,
"step": 571
},
{
"epoch": 0.6992665036674817,
"grad_norm": 0.41970741748809814,
"learning_rate": 1.4567259365060738e-06,
"loss": 0.026068419218063354,
"step": 572
},
{
"epoch": 0.7004889975550123,
"grad_norm": 0.7427928447723389,
"learning_rate": 1.4550167921427218e-06,
"loss": 0.05480552092194557,
"step": 573
},
{
"epoch": 0.7017114914425427,
"grad_norm": 0.8021201491355896,
"learning_rate": 1.4533059699011464e-06,
"loss": 0.10090692341327667,
"step": 574
},
{
"epoch": 0.7029339853300733,
"grad_norm": 1.0930843353271484,
"learning_rate": 1.4515934760900183e-06,
"loss": 0.1564808189868927,
"step": 575
},
{
"epoch": 0.7041564792176039,
"grad_norm": 0.5146190524101257,
"learning_rate": 1.4498793170241734e-06,
"loss": 0.044559553265571594,
"step": 576
},
{
"epoch": 0.7053789731051344,
"grad_norm": 0.5048078894615173,
"learning_rate": 1.448163499024587e-06,
"loss": 0.03638163208961487,
"step": 577
},
{
"epoch": 0.706601466992665,
"grad_norm": 0.39478570222854614,
"learning_rate": 1.4464460284183525e-06,
"loss": 0.02867947146296501,
"step": 578
},
{
"epoch": 0.7078239608801956,
"grad_norm": 0.5669252276420593,
"learning_rate": 1.444726911538657e-06,
"loss": 0.06127239018678665,
"step": 579
},
{
"epoch": 0.7090464547677262,
"grad_norm": 0.4017905294895172,
"learning_rate": 1.4430061547247587e-06,
"loss": 0.0282396599650383,
"step": 580
},
{
"epoch": 0.7102689486552567,
"grad_norm": 0.9176173210144043,
"learning_rate": 1.4412837643219624e-06,
"loss": 0.06624332815408707,
"step": 581
},
{
"epoch": 0.7114914425427873,
"grad_norm": 0.6966966390609741,
"learning_rate": 1.4395597466815975e-06,
"loss": 0.08339723944664001,
"step": 582
},
{
"epoch": 0.7127139364303179,
"grad_norm": 2.1000962257385254,
"learning_rate": 1.4378341081609927e-06,
"loss": 0.17328910529613495,
"step": 583
},
{
"epoch": 0.7139364303178484,
"grad_norm": 0.47194063663482666,
"learning_rate": 1.4361068551234554e-06,
"loss": 0.027584604918956757,
"step": 584
},
{
"epoch": 0.715158924205379,
"grad_norm": 0.43176713585853577,
"learning_rate": 1.434377993938245e-06,
"loss": 0.03905326500535011,
"step": 585
},
{
"epoch": 0.7163814180929096,
"grad_norm": 0.6381475925445557,
"learning_rate": 1.4326475309805517e-06,
"loss": 0.06634974479675293,
"step": 586
},
{
"epoch": 0.71760391198044,
"grad_norm": 0.793116569519043,
"learning_rate": 1.430915472631472e-06,
"loss": 0.12857578694820404,
"step": 587
},
{
"epoch": 0.7188264058679706,
"grad_norm": 0.4816029369831085,
"learning_rate": 1.429181825277986e-06,
"loss": 0.03482411056756973,
"step": 588
},
{
"epoch": 0.7200488997555012,
"grad_norm": 0.5482797026634216,
"learning_rate": 1.4274465953129325e-06,
"loss": 0.061665598303079605,
"step": 589
},
{
"epoch": 0.7212713936430318,
"grad_norm": 0.6367279887199402,
"learning_rate": 1.4257097891349861e-06,
"loss": 0.07600852102041245,
"step": 590
},
{
"epoch": 0.7224938875305623,
"grad_norm": 0.3412401080131531,
"learning_rate": 1.4239714131486346e-06,
"loss": 0.025902170687913895,
"step": 591
},
{
"epoch": 0.7237163814180929,
"grad_norm": 0.6624696850776672,
"learning_rate": 1.4222314737641542e-06,
"loss": 0.09334467351436615,
"step": 592
},
{
"epoch": 0.7249388753056235,
"grad_norm": 0.4818311035633087,
"learning_rate": 1.4204899773975854e-06,
"loss": 0.04154723137617111,
"step": 593
},
{
"epoch": 0.726161369193154,
"grad_norm": 0.9077998995780945,
"learning_rate": 1.4187469304707107e-06,
"loss": 0.17066936194896698,
"step": 594
},
{
"epoch": 0.7273838630806846,
"grad_norm": 0.6245238184928894,
"learning_rate": 1.4170023394110304e-06,
"loss": 0.06443880498409271,
"step": 595
},
{
"epoch": 0.7286063569682152,
"grad_norm": 0.5550680756568909,
"learning_rate": 1.4152562106517385e-06,
"loss": 0.031198600307106972,
"step": 596
},
{
"epoch": 0.7298288508557457,
"grad_norm": 0.549848198890686,
"learning_rate": 1.4135085506316996e-06,
"loss": 0.045275889337062836,
"step": 597
},
{
"epoch": 0.7310513447432763,
"grad_norm": 1.5673190355300903,
"learning_rate": 1.4117593657954241e-06,
"loss": 0.16425545513629913,
"step": 598
},
{
"epoch": 0.7322738386308069,
"grad_norm": 1.1526298522949219,
"learning_rate": 1.4100086625930462e-06,
"loss": 0.22205528616905212,
"step": 599
},
{
"epoch": 0.7334963325183375,
"grad_norm": 0.7212430834770203,
"learning_rate": 1.4082564474802984e-06,
"loss": 0.06007983162999153,
"step": 600
},
{
"epoch": 0.7347188264058679,
"grad_norm": 0.9007821083068848,
"learning_rate": 1.4065027269184886e-06,
"loss": 0.0697171688079834,
"step": 601
},
{
"epoch": 0.7359413202933985,
"grad_norm": 0.7349361777305603,
"learning_rate": 1.4047475073744762e-06,
"loss": 0.09240229427814484,
"step": 602
},
{
"epoch": 0.7371638141809291,
"grad_norm": 0.4853475093841553,
"learning_rate": 1.4029907953206474e-06,
"loss": 0.049861498177051544,
"step": 603
},
{
"epoch": 0.7383863080684596,
"grad_norm": 0.7557780742645264,
"learning_rate": 1.4012325972348933e-06,
"loss": 0.10245846956968307,
"step": 604
},
{
"epoch": 0.7396088019559902,
"grad_norm": 0.8108947277069092,
"learning_rate": 1.3994729196005837e-06,
"loss": 0.08864305913448334,
"step": 605
},
{
"epoch": 0.7408312958435208,
"grad_norm": 0.6105418801307678,
"learning_rate": 1.3977117689065443e-06,
"loss": 0.06996094435453415,
"step": 606
},
{
"epoch": 0.7420537897310513,
"grad_norm": 0.7286546230316162,
"learning_rate": 1.3959491516470332e-06,
"loss": 0.097049281001091,
"step": 607
},
{
"epoch": 0.7432762836185819,
"grad_norm": 0.4704604148864746,
"learning_rate": 1.3941850743217167e-06,
"loss": 0.03422698751091957,
"step": 608
},
{
"epoch": 0.7444987775061125,
"grad_norm": 0.8591526746749878,
"learning_rate": 1.3924195434356441e-06,
"loss": 0.08008885383605957,
"step": 609
},
{
"epoch": 0.7457212713936431,
"grad_norm": 0.6949493885040283,
"learning_rate": 1.3906525654992254e-06,
"loss": 0.07364436984062195,
"step": 610
},
{
"epoch": 0.7469437652811736,
"grad_norm": 0.7066441178321838,
"learning_rate": 1.388884147028207e-06,
"loss": 0.07071486115455627,
"step": 611
},
{
"epoch": 0.7481662591687042,
"grad_norm": 0.7446351647377014,
"learning_rate": 1.3871142945436463e-06,
"loss": 0.06419433653354645,
"step": 612
},
{
"epoch": 0.7493887530562348,
"grad_norm": 0.41348668932914734,
"learning_rate": 1.385343014571889e-06,
"loss": 0.029183046892285347,
"step": 613
},
{
"epoch": 0.7506112469437652,
"grad_norm": 0.5886424779891968,
"learning_rate": 1.383570313644545e-06,
"loss": 0.05191269516944885,
"step": 614
},
{
"epoch": 0.7518337408312958,
"grad_norm": 0.5245025157928467,
"learning_rate": 1.3817961982984644e-06,
"loss": 0.03043348714709282,
"step": 615
},
{
"epoch": 0.7530562347188264,
"grad_norm": 0.5730315446853638,
"learning_rate": 1.3800206750757113e-06,
"loss": 0.04677876457571983,
"step": 616
},
{
"epoch": 0.7542787286063569,
"grad_norm": 0.5601646900177002,
"learning_rate": 1.378243750523543e-06,
"loss": 0.04672516882419586,
"step": 617
},
{
"epoch": 0.7555012224938875,
"grad_norm": 0.3911654055118561,
"learning_rate": 1.3764654311943838e-06,
"loss": 0.02169198729097843,
"step": 618
},
{
"epoch": 0.7567237163814181,
"grad_norm": 0.5997183918952942,
"learning_rate": 1.3746857236458005e-06,
"loss": 0.05103592574596405,
"step": 619
},
{
"epoch": 0.7579462102689487,
"grad_norm": 0.7790828347206116,
"learning_rate": 1.3729046344404799e-06,
"loss": 0.0953461080789566,
"step": 620
},
{
"epoch": 0.7591687041564792,
"grad_norm": 0.6592403650283813,
"learning_rate": 1.3711221701462034e-06,
"loss": 0.06809133291244507,
"step": 621
},
{
"epoch": 0.7603911980440098,
"grad_norm": 0.5518736243247986,
"learning_rate": 1.3693383373358232e-06,
"loss": 0.05714889243245125,
"step": 622
},
{
"epoch": 0.7616136919315404,
"grad_norm": 0.6428724527359009,
"learning_rate": 1.367553142587237e-06,
"loss": 0.07214516401290894,
"step": 623
},
{
"epoch": 0.7628361858190709,
"grad_norm": 0.6274532675743103,
"learning_rate": 1.3657665924833662e-06,
"loss": 0.059236541390419006,
"step": 624
},
{
"epoch": 0.7640586797066015,
"grad_norm": 0.8565517067909241,
"learning_rate": 1.3639786936121286e-06,
"loss": 0.10701166838407516,
"step": 625
},
{
"epoch": 0.7652811735941321,
"grad_norm": 0.6508960127830505,
"learning_rate": 1.3621894525664161e-06,
"loss": 0.06948018074035645,
"step": 626
},
{
"epoch": 0.7665036674816625,
"grad_norm": 0.8504416346549988,
"learning_rate": 1.3603988759440707e-06,
"loss": 0.08728962391614914,
"step": 627
},
{
"epoch": 0.7677261613691931,
"grad_norm": 0.8560598492622375,
"learning_rate": 1.3586069703478585e-06,
"loss": 0.14345481991767883,
"step": 628
},
{
"epoch": 0.7689486552567237,
"grad_norm": 1.6578972339630127,
"learning_rate": 1.3568137423854457e-06,
"loss": 0.1032346859574318,
"step": 629
},
{
"epoch": 0.7701711491442543,
"grad_norm": 0.5199538469314575,
"learning_rate": 1.355019198669376e-06,
"loss": 0.03024023026227951,
"step": 630
},
{
"epoch": 0.7713936430317848,
"grad_norm": 0.8251024484634399,
"learning_rate": 1.3532233458170443e-06,
"loss": 0.07182576507329941,
"step": 631
},
{
"epoch": 0.7726161369193154,
"grad_norm": 0.45426490902900696,
"learning_rate": 1.351426190450673e-06,
"loss": 0.03570329397916794,
"step": 632
},
{
"epoch": 0.773838630806846,
"grad_norm": 0.35087963938713074,
"learning_rate": 1.3496277391972872e-06,
"loss": 0.016510942950844765,
"step": 633
},
{
"epoch": 0.7750611246943765,
"grad_norm": 0.9418892860412598,
"learning_rate": 1.3478279986886913e-06,
"loss": 0.10918392241001129,
"step": 634
},
{
"epoch": 0.7762836185819071,
"grad_norm": 0.8411474227905273,
"learning_rate": 1.3460269755614435e-06,
"loss": 0.093037910759449,
"step": 635
},
{
"epoch": 0.7775061124694377,
"grad_norm": 0.42371276021003723,
"learning_rate": 1.3442246764568313e-06,
"loss": 0.028708595782518387,
"step": 636
},
{
"epoch": 0.7787286063569682,
"grad_norm": 0.4758448898792267,
"learning_rate": 1.3424211080208476e-06,
"loss": 0.023815512657165527,
"step": 637
},
{
"epoch": 0.7799511002444988,
"grad_norm": 0.6919294595718384,
"learning_rate": 1.340616276904167e-06,
"loss": 0.07408645004034042,
"step": 638
},
{
"epoch": 0.7811735941320294,
"grad_norm": 0.7298597097396851,
"learning_rate": 1.3388101897621183e-06,
"loss": 0.10057426244020462,
"step": 639
},
{
"epoch": 0.78239608801956,
"grad_norm": 0.2879655361175537,
"learning_rate": 1.3370028532546634e-06,
"loss": 0.03451251983642578,
"step": 640
},
{
"epoch": 0.7836185819070904,
"grad_norm": 1.6086695194244385,
"learning_rate": 1.3351942740463703e-06,
"loss": 0.1344827115535736,
"step": 641
},
{
"epoch": 0.784841075794621,
"grad_norm": 0.4902389943599701,
"learning_rate": 1.333384458806391e-06,
"loss": 0.0572238489985466,
"step": 642
},
{
"epoch": 0.7860635696821516,
"grad_norm": 0.6465232372283936,
"learning_rate": 1.3315734142084334e-06,
"loss": 0.07329805195331573,
"step": 643
},
{
"epoch": 0.7872860635696821,
"grad_norm": 0.7691776752471924,
"learning_rate": 1.3297611469307398e-06,
"loss": 0.09008753299713135,
"step": 644
},
{
"epoch": 0.7885085574572127,
"grad_norm": 0.6296072006225586,
"learning_rate": 1.3279476636560607e-06,
"loss": 0.06155703589320183,
"step": 645
},
{
"epoch": 0.7897310513447433,
"grad_norm": 1.0899748802185059,
"learning_rate": 1.3261329710716316e-06,
"loss": 0.16210412979125977,
"step": 646
},
{
"epoch": 0.7909535452322738,
"grad_norm": 1.4416887760162354,
"learning_rate": 1.3243170758691458e-06,
"loss": 0.1272645890712738,
"step": 647
},
{
"epoch": 0.7921760391198044,
"grad_norm": 0.9231910109519958,
"learning_rate": 1.3224999847447325e-06,
"loss": 0.10777764767408371,
"step": 648
},
{
"epoch": 0.793398533007335,
"grad_norm": 0.8785171508789062,
"learning_rate": 1.32068170439893e-06,
"loss": 0.11167420446872711,
"step": 649
},
{
"epoch": 0.7946210268948656,
"grad_norm": 0.5500822067260742,
"learning_rate": 1.3188622415366628e-06,
"loss": 0.05699105188250542,
"step": 650
},
{
"epoch": 0.7958435207823961,
"grad_norm": 1.3321093320846558,
"learning_rate": 1.317041602867215e-06,
"loss": 0.07355312258005142,
"step": 651
},
{
"epoch": 0.7970660146699267,
"grad_norm": 1.7610746622085571,
"learning_rate": 1.3152197951042067e-06,
"loss": 0.1155061274766922,
"step": 652
},
{
"epoch": 0.7982885085574573,
"grad_norm": 0.46623730659484863,
"learning_rate": 1.31339682496557e-06,
"loss": 0.045190226286649704,
"step": 653
},
{
"epoch": 0.7995110024449877,
"grad_norm": 0.9527392983436584,
"learning_rate": 1.3115726991735221e-06,
"loss": 0.07963813096284866,
"step": 654
},
{
"epoch": 0.8007334963325183,
"grad_norm": 0.7273218035697937,
"learning_rate": 1.3097474244545418e-06,
"loss": 0.04045310989022255,
"step": 655
},
{
"epoch": 0.8019559902200489,
"grad_norm": 0.43269044160842896,
"learning_rate": 1.3079210075393453e-06,
"loss": 0.04177930951118469,
"step": 656
},
{
"epoch": 0.8031784841075794,
"grad_norm": 0.838444173336029,
"learning_rate": 1.3060934551628602e-06,
"loss": 0.07866289466619492,
"step": 657
},
{
"epoch": 0.80440097799511,
"grad_norm": 0.44775280356407166,
"learning_rate": 1.304264774064201e-06,
"loss": 0.03439653664827347,
"step": 658
},
{
"epoch": 0.8056234718826406,
"grad_norm": 0.7083353400230408,
"learning_rate": 1.3024349709866448e-06,
"loss": 0.06291119754314423,
"step": 659
},
{
"epoch": 0.8068459657701712,
"grad_norm": 0.49827396869659424,
"learning_rate": 1.3006040526776051e-06,
"loss": 0.03395054116845131,
"step": 660
},
{
"epoch": 0.8080684596577017,
"grad_norm": 1.672579050064087,
"learning_rate": 1.2987720258886092e-06,
"loss": 0.11172181367874146,
"step": 661
},
{
"epoch": 0.8092909535452323,
"grad_norm": 0.6045190095901489,
"learning_rate": 1.2969388973752712e-06,
"loss": 0.06113462895154953,
"step": 662
},
{
"epoch": 0.8105134474327629,
"grad_norm": 0.606536865234375,
"learning_rate": 1.295104673897267e-06,
"loss": 0.030633430927991867,
"step": 663
},
{
"epoch": 0.8117359413202934,
"grad_norm": 0.4533053934574127,
"learning_rate": 1.2932693622183117e-06,
"loss": 0.03425963222980499,
"step": 664
},
{
"epoch": 0.812958435207824,
"grad_norm": 0.46414250135421753,
"learning_rate": 1.2914329691061325e-06,
"loss": 0.04182780534029007,
"step": 665
},
{
"epoch": 0.8141809290953546,
"grad_norm": 0.7817361354827881,
"learning_rate": 1.289595501332444e-06,
"loss": 0.09963971376419067,
"step": 666
},
{
"epoch": 0.815403422982885,
"grad_norm": 0.6154112219810486,
"learning_rate": 1.287756965672924e-06,
"loss": 0.055649533867836,
"step": 667
},
{
"epoch": 0.8166259168704156,
"grad_norm": 0.5807240605354309,
"learning_rate": 1.2859173689071885e-06,
"loss": 0.06141240522265434,
"step": 668
},
{
"epoch": 0.8178484107579462,
"grad_norm": 3.559547185897827,
"learning_rate": 1.2840767178187654e-06,
"loss": 0.12596261501312256,
"step": 669
},
{
"epoch": 0.8190709046454768,
"grad_norm": 0.5067200064659119,
"learning_rate": 1.2822350191950713e-06,
"loss": 0.035477496683597565,
"step": 670
},
{
"epoch": 0.8202933985330073,
"grad_norm": 2.064934015274048,
"learning_rate": 1.2803922798273853e-06,
"loss": 0.1219458281993866,
"step": 671
},
{
"epoch": 0.8215158924205379,
"grad_norm": 0.4978325664997101,
"learning_rate": 1.278548506510824e-06,
"loss": 0.04616785794496536,
"step": 672
},
{
"epoch": 0.8227383863080685,
"grad_norm": 0.8238120675086975,
"learning_rate": 1.276703706044317e-06,
"loss": 0.09705391526222229,
"step": 673
},
{
"epoch": 0.823960880195599,
"grad_norm": 0.6087835431098938,
"learning_rate": 1.2748578852305815e-06,
"loss": 0.05570684373378754,
"step": 674
},
{
"epoch": 0.8251833740831296,
"grad_norm": 0.504554033279419,
"learning_rate": 1.2730110508760974e-06,
"loss": 0.03094184771180153,
"step": 675
},
{
"epoch": 0.8264058679706602,
"grad_norm": 0.6048548221588135,
"learning_rate": 1.2711632097910812e-06,
"loss": 0.054385650902986526,
"step": 676
},
{
"epoch": 0.8276283618581907,
"grad_norm": 0.7428304553031921,
"learning_rate": 1.2693143687894628e-06,
"loss": 0.06875768303871155,
"step": 677
},
{
"epoch": 0.8288508557457213,
"grad_norm": 0.5322964191436768,
"learning_rate": 1.2674645346888589e-06,
"loss": 0.05431134253740311,
"step": 678
},
{
"epoch": 0.8300733496332519,
"grad_norm": 0.537516176700592,
"learning_rate": 1.265613714310548e-06,
"loss": 0.044658176600933075,
"step": 679
},
{
"epoch": 0.8312958435207825,
"grad_norm": 1.4021778106689453,
"learning_rate": 1.2637619144794456e-06,
"loss": 0.0786314383149147,
"step": 680
},
{
"epoch": 0.8325183374083129,
"grad_norm": 0.5931991934776306,
"learning_rate": 1.2619091420240793e-06,
"loss": 0.08273275196552277,
"step": 681
},
{
"epoch": 0.8337408312958435,
"grad_norm": 1.9092599153518677,
"learning_rate": 1.2600554037765631e-06,
"loss": 0.15886518359184265,
"step": 682
},
{
"epoch": 0.8349633251833741,
"grad_norm": 0.4309159517288208,
"learning_rate": 1.258200706572572e-06,
"loss": 0.04493548348546028,
"step": 683
},
{
"epoch": 0.8361858190709046,
"grad_norm": 0.7797632217407227,
"learning_rate": 1.2563450572513174e-06,
"loss": 0.05439355969429016,
"step": 684
},
{
"epoch": 0.8374083129584352,
"grad_norm": 0.7305620908737183,
"learning_rate": 1.2544884626555223e-06,
"loss": 0.08348140120506287,
"step": 685
},
{
"epoch": 0.8386308068459658,
"grad_norm": 0.5510923862457275,
"learning_rate": 1.252630929631394e-06,
"loss": 0.03356444463133812,
"step": 686
},
{
"epoch": 0.8398533007334963,
"grad_norm": 0.5874477028846741,
"learning_rate": 1.2507724650286012e-06,
"loss": 0.06791111826896667,
"step": 687
},
{
"epoch": 0.8410757946210269,
"grad_norm": 0.6202130913734436,
"learning_rate": 1.2489130757002482e-06,
"loss": 0.05181968957185745,
"step": 688
},
{
"epoch": 0.8422982885085575,
"grad_norm": 0.8710898756980896,
"learning_rate": 1.2470527685028482e-06,
"loss": 0.11042847484350204,
"step": 689
},
{
"epoch": 0.843520782396088,
"grad_norm": 0.7173504829406738,
"learning_rate": 1.2451915502962995e-06,
"loss": 0.06784708797931671,
"step": 690
},
{
"epoch": 0.8447432762836186,
"grad_norm": 0.4389496445655823,
"learning_rate": 1.2433294279438602e-06,
"loss": 0.03387697413563728,
"step": 691
},
{
"epoch": 0.8459657701711492,
"grad_norm": 0.5625354647636414,
"learning_rate": 1.2414664083121215e-06,
"loss": 0.03610862046480179,
"step": 692
},
{
"epoch": 0.8471882640586798,
"grad_norm": 0.5940917730331421,
"learning_rate": 1.2396024982709844e-06,
"loss": 0.05663711950182915,
"step": 693
},
{
"epoch": 0.8484107579462102,
"grad_norm": 0.8912255167961121,
"learning_rate": 1.2377377046936323e-06,
"loss": 0.09585443139076233,
"step": 694
},
{
"epoch": 0.8496332518337408,
"grad_norm": 0.7051909565925598,
"learning_rate": 1.2358720344565074e-06,
"loss": 0.04279647767543793,
"step": 695
},
{
"epoch": 0.8508557457212714,
"grad_norm": 0.9704980850219727,
"learning_rate": 1.2340054944392842e-06,
"loss": 0.13391537964344025,
"step": 696
},
{
"epoch": 0.8520782396088019,
"grad_norm": 0.3952852785587311,
"learning_rate": 1.2321380915248445e-06,
"loss": 0.028005288913846016,
"step": 697
},
{
"epoch": 0.8533007334963325,
"grad_norm": 0.6785120368003845,
"learning_rate": 1.2302698325992525e-06,
"loss": 0.07685574889183044,
"step": 698
},
{
"epoch": 0.8545232273838631,
"grad_norm": 0.5275679230690002,
"learning_rate": 1.228400724551728e-06,
"loss": 0.041168440133333206,
"step": 699
},
{
"epoch": 0.8557457212713936,
"grad_norm": 0.42105647921562195,
"learning_rate": 1.2265307742746232e-06,
"loss": 0.02345583215355873,
"step": 700
},
{
"epoch": 0.8569682151589242,
"grad_norm": 0.9160144925117493,
"learning_rate": 1.224659988663395e-06,
"loss": 0.09545909613370895,
"step": 701
},
{
"epoch": 0.8581907090464548,
"grad_norm": 1.097485899925232,
"learning_rate": 1.2227883746165813e-06,
"loss": 0.08808273077011108,
"step": 702
},
{
"epoch": 0.8594132029339854,
"grad_norm": 0.5029830932617188,
"learning_rate": 1.2209159390357737e-06,
"loss": 0.06424691528081894,
"step": 703
},
{
"epoch": 0.8606356968215159,
"grad_norm": 0.6898568272590637,
"learning_rate": 1.2190426888255951e-06,
"loss": 0.06333495676517487,
"step": 704
},
{
"epoch": 0.8618581907090465,
"grad_norm": 0.5062806606292725,
"learning_rate": 1.2171686308936709e-06,
"loss": 0.04386819899082184,
"step": 705
},
{
"epoch": 0.863080684596577,
"grad_norm": 1.5881450176239014,
"learning_rate": 1.215293772150605e-06,
"loss": 0.14586570858955383,
"step": 706
},
{
"epoch": 0.8643031784841075,
"grad_norm": 0.5778710246086121,
"learning_rate": 1.2134181195099554e-06,
"loss": 0.06586824357509613,
"step": 707
},
{
"epoch": 0.8655256723716381,
"grad_norm": 0.5900859236717224,
"learning_rate": 1.2115416798882062e-06,
"loss": 0.06384717673063278,
"step": 708
},
{
"epoch": 0.8667481662591687,
"grad_norm": 0.6273130178451538,
"learning_rate": 1.2096644602047445e-06,
"loss": 0.07862474024295807,
"step": 709
},
{
"epoch": 0.8679706601466992,
"grad_norm": 0.5142886638641357,
"learning_rate": 1.2077864673818338e-06,
"loss": 0.03151443600654602,
"step": 710
},
{
"epoch": 0.8691931540342298,
"grad_norm": 0.7205119729042053,
"learning_rate": 1.2059077083445878e-06,
"loss": 0.0819907858967781,
"step": 711
},
{
"epoch": 0.8704156479217604,
"grad_norm": 0.5802266597747803,
"learning_rate": 1.204028190020947e-06,
"loss": 0.040982816368341446,
"step": 712
},
{
"epoch": 0.871638141809291,
"grad_norm": 0.4787074625492096,
"learning_rate": 1.2021479193416499e-06,
"loss": 0.026309026405215263,
"step": 713
},
{
"epoch": 0.8728606356968215,
"grad_norm": 0.421750545501709,
"learning_rate": 1.2002669032402108e-06,
"loss": 0.027960505336523056,
"step": 714
},
{
"epoch": 0.8740831295843521,
"grad_norm": 0.5299254655838013,
"learning_rate": 1.1983851486528925e-06,
"loss": 0.05148579552769661,
"step": 715
},
{
"epoch": 0.8753056234718827,
"grad_norm": 0.7139096856117249,
"learning_rate": 1.1965026625186804e-06,
"loss": 0.07582290470600128,
"step": 716
},
{
"epoch": 0.8765281173594132,
"grad_norm": 0.8392231464385986,
"learning_rate": 1.1946194517792584e-06,
"loss": 0.07878116518259048,
"step": 717
},
{
"epoch": 0.8777506112469438,
"grad_norm": 0.7150721549987793,
"learning_rate": 1.192735523378981e-06,
"loss": 0.061754804104566574,
"step": 718
},
{
"epoch": 0.8789731051344744,
"grad_norm": 0.45794108510017395,
"learning_rate": 1.1908508842648505e-06,
"loss": 0.0430489107966423,
"step": 719
},
{
"epoch": 0.8801955990220048,
"grad_norm": 0.5038924217224121,
"learning_rate": 1.1889655413864897e-06,
"loss": 0.026686066761612892,
"step": 720
},
{
"epoch": 0.8814180929095354,
"grad_norm": 0.43425583839416504,
"learning_rate": 1.1870795016961155e-06,
"loss": 0.023565402254462242,
"step": 721
},
{
"epoch": 0.882640586797066,
"grad_norm": 0.45350778102874756,
"learning_rate": 1.185192772148515e-06,
"loss": 0.03702652081847191,
"step": 722
},
{
"epoch": 0.8838630806845966,
"grad_norm": 1.0698463916778564,
"learning_rate": 1.1833053597010202e-06,
"loss": 0.12560996413230896,
"step": 723
},
{
"epoch": 0.8850855745721271,
"grad_norm": 0.5621904730796814,
"learning_rate": 1.181417271313479e-06,
"loss": 0.05158263072371483,
"step": 724
},
{
"epoch": 0.8863080684596577,
"grad_norm": 0.5964070558547974,
"learning_rate": 1.179528513948234e-06,
"loss": 0.059765592217445374,
"step": 725
},
{
"epoch": 0.8875305623471883,
"grad_norm": 0.5131673812866211,
"learning_rate": 1.177639094570093e-06,
"loss": 0.0434177964925766,
"step": 726
},
{
"epoch": 0.8887530562347188,
"grad_norm": 0.617845356464386,
"learning_rate": 1.1757490201463065e-06,
"loss": 0.05792270228266716,
"step": 727
},
{
"epoch": 0.8899755501222494,
"grad_norm": 0.780532956123352,
"learning_rate": 1.1738582976465397e-06,
"loss": 0.09840493649244308,
"step": 728
},
{
"epoch": 0.89119804400978,
"grad_norm": 0.6025611758232117,
"learning_rate": 1.171966934042847e-06,
"loss": 0.03451397642493248,
"step": 729
},
{
"epoch": 0.8924205378973105,
"grad_norm": 0.724324107170105,
"learning_rate": 1.1700749363096479e-06,
"loss": 0.10366087406873703,
"step": 730
},
{
"epoch": 0.8936430317848411,
"grad_norm": 0.5929924845695496,
"learning_rate": 1.1681823114236998e-06,
"loss": 0.03056013584136963,
"step": 731
},
{
"epoch": 0.8948655256723717,
"grad_norm": 1.0835386514663696,
"learning_rate": 1.1662890663640729e-06,
"loss": 0.14992627501487732,
"step": 732
},
{
"epoch": 0.8960880195599022,
"grad_norm": 0.8938685059547424,
"learning_rate": 1.1643952081121239e-06,
"loss": 0.1557403951883316,
"step": 733
},
{
"epoch": 0.8973105134474327,
"grad_norm": 0.49020227789878845,
"learning_rate": 1.1625007436514712e-06,
"loss": 0.03596637025475502,
"step": 734
},
{
"epoch": 0.8985330073349633,
"grad_norm": 0.667955756187439,
"learning_rate": 1.1606056799679682e-06,
"loss": 0.05800396203994751,
"step": 735
},
{
"epoch": 0.8997555012224939,
"grad_norm": 0.8645825386047363,
"learning_rate": 1.1587100240496778e-06,
"loss": 0.1001749336719513,
"step": 736
},
{
"epoch": 0.9009779951100244,
"grad_norm": 0.5881163477897644,
"learning_rate": 1.1568137828868476e-06,
"loss": 0.06772617995738983,
"step": 737
},
{
"epoch": 0.902200488997555,
"grad_norm": 0.6839173436164856,
"learning_rate": 1.1549169634718827e-06,
"loss": 0.11450149118900299,
"step": 738
},
{
"epoch": 0.9034229828850856,
"grad_norm": 0.7191789746284485,
"learning_rate": 1.1530195727993199e-06,
"loss": 0.08821175247430801,
"step": 739
},
{
"epoch": 0.9046454767726161,
"grad_norm": 0.6566911339759827,
"learning_rate": 1.1511216178658037e-06,
"loss": 0.04440095275640488,
"step": 740
},
{
"epoch": 0.9058679706601467,
"grad_norm": 0.4701736271381378,
"learning_rate": 1.149223105670059e-06,
"loss": 0.05393720045685768,
"step": 741
},
{
"epoch": 0.9070904645476773,
"grad_norm": 0.7630343437194824,
"learning_rate": 1.1473240432128645e-06,
"loss": 0.09228914231061935,
"step": 742
},
{
"epoch": 0.9083129584352079,
"grad_norm": 0.49489596486091614,
"learning_rate": 1.1454244374970297e-06,
"loss": 0.04277567192912102,
"step": 743
},
{
"epoch": 0.9095354523227384,
"grad_norm": 1.150696039199829,
"learning_rate": 1.143524295527366e-06,
"loss": 0.2206958383321762,
"step": 744
},
{
"epoch": 0.910757946210269,
"grad_norm": 0.46509072184562683,
"learning_rate": 1.1416236243106637e-06,
"loss": 0.02937270887196064,
"step": 745
},
{
"epoch": 0.9119804400977995,
"grad_norm": 0.5618892908096313,
"learning_rate": 1.1397224308556626e-06,
"loss": 0.06564927101135254,
"step": 746
},
{
"epoch": 0.91320293398533,
"grad_norm": 0.34354090690612793,
"learning_rate": 1.1378207221730301e-06,
"loss": 0.027808433398604393,
"step": 747
},
{
"epoch": 0.9144254278728606,
"grad_norm": 0.44182536005973816,
"learning_rate": 1.135918505275333e-06,
"loss": 0.031185872852802277,
"step": 748
},
{
"epoch": 0.9156479217603912,
"grad_norm": 0.9432840347290039,
"learning_rate": 1.1340157871770115e-06,
"loss": 0.0698775202035904,
"step": 749
},
{
"epoch": 0.9168704156479217,
"grad_norm": 0.5057671666145325,
"learning_rate": 1.132112574894355e-06,
"loss": 0.04757803678512573,
"step": 750
},
{
"epoch": 0.9180929095354523,
"grad_norm": 0.47517630457878113,
"learning_rate": 1.1302088754454744e-06,
"loss": 0.03897363319993019,
"step": 751
},
{
"epoch": 0.9193154034229829,
"grad_norm": 1.6351253986358643,
"learning_rate": 1.1283046958502776e-06,
"loss": 0.13018938899040222,
"step": 752
},
{
"epoch": 0.9205378973105135,
"grad_norm": 0.3650830090045929,
"learning_rate": 1.1264000431304422e-06,
"loss": 0.033754028379917145,
"step": 753
},
{
"epoch": 0.921760391198044,
"grad_norm": 0.642149806022644,
"learning_rate": 1.1244949243093915e-06,
"loss": 0.04887555167078972,
"step": 754
},
{
"epoch": 0.9229828850855746,
"grad_norm": 0.8852201700210571,
"learning_rate": 1.1225893464122672e-06,
"loss": 0.1408361792564392,
"step": 755
},
{
"epoch": 0.9242053789731052,
"grad_norm": 0.4289619028568268,
"learning_rate": 1.1206833164659032e-06,
"loss": 0.029355429112911224,
"step": 756
},
{
"epoch": 0.9254278728606357,
"grad_norm": 0.7418862581253052,
"learning_rate": 1.1187768414988013e-06,
"loss": 0.09785927832126617,
"step": 757
},
{
"epoch": 0.9266503667481663,
"grad_norm": 0.7222095727920532,
"learning_rate": 1.116869928541104e-06,
"loss": 0.08119559288024902,
"step": 758
},
{
"epoch": 0.9278728606356969,
"grad_norm": 1.0154494047164917,
"learning_rate": 1.1149625846245681e-06,
"loss": 0.12039929628372192,
"step": 759
},
{
"epoch": 0.9290953545232273,
"grad_norm": 0.5429635643959045,
"learning_rate": 1.1130548167825407e-06,
"loss": 0.056033652275800705,
"step": 760
},
{
"epoch": 0.9303178484107579,
"grad_norm": 0.5572449564933777,
"learning_rate": 1.1111466320499315e-06,
"loss": 0.04957162216305733,
"step": 761
},
{
"epoch": 0.9315403422982885,
"grad_norm": 0.3672495186328888,
"learning_rate": 1.1092380374631878e-06,
"loss": 0.031167401000857353,
"step": 762
},
{
"epoch": 0.9327628361858191,
"grad_norm": 0.6291825771331787,
"learning_rate": 1.107329040060268e-06,
"loss": 0.05553711950778961,
"step": 763
},
{
"epoch": 0.9339853300733496,
"grad_norm": 0.7674726843833923,
"learning_rate": 1.105419646880616e-06,
"loss": 0.05721331760287285,
"step": 764
},
{
"epoch": 0.9352078239608802,
"grad_norm": 0.7119777798652649,
"learning_rate": 1.1035098649651353e-06,
"loss": 0.0921907052397728,
"step": 765
},
{
"epoch": 0.9364303178484108,
"grad_norm": 0.5140713453292847,
"learning_rate": 1.1015997013561624e-06,
"loss": 0.05349583923816681,
"step": 766
},
{
"epoch": 0.9376528117359413,
"grad_norm": 0.40966829657554626,
"learning_rate": 1.0996891630974414e-06,
"loss": 0.020671812817454338,
"step": 767
},
{
"epoch": 0.9388753056234719,
"grad_norm": 0.3428141176700592,
"learning_rate": 1.0977782572340988e-06,
"loss": 0.01985091157257557,
"step": 768
},
{
"epoch": 0.9400977995110025,
"grad_norm": 0.48155471682548523,
"learning_rate": 1.095866990812615e-06,
"loss": 0.025409545749425888,
"step": 769
},
{
"epoch": 0.941320293398533,
"grad_norm": 0.5641920566558838,
"learning_rate": 1.0939553708808015e-06,
"loss": 0.057349830865859985,
"step": 770
},
{
"epoch": 0.9425427872860636,
"grad_norm": 0.6109334230422974,
"learning_rate": 1.092043404487773e-06,
"loss": 0.0462774783372879,
"step": 771
},
{
"epoch": 0.9437652811735942,
"grad_norm": 0.679800271987915,
"learning_rate": 1.090131098683921e-06,
"loss": 0.06346183270215988,
"step": 772
},
{
"epoch": 0.9449877750611247,
"grad_norm": 0.7987798452377319,
"learning_rate": 1.0882184605208893e-06,
"loss": 0.044133640825748444,
"step": 773
},
{
"epoch": 0.9462102689486552,
"grad_norm": 0.6996093392372131,
"learning_rate": 1.0863054970515473e-06,
"loss": 0.08346416801214218,
"step": 774
},
{
"epoch": 0.9474327628361858,
"grad_norm": 0.5736404061317444,
"learning_rate": 1.0843922153299636e-06,
"loss": 0.049279890954494476,
"step": 775
},
{
"epoch": 0.9486552567237164,
"grad_norm": 0.39150163531303406,
"learning_rate": 1.0824786224113804e-06,
"loss": 0.025282911956310272,
"step": 776
},
{
"epoch": 0.9498777506112469,
"grad_norm": 0.6770985722541809,
"learning_rate": 1.080564725352188e-06,
"loss": 0.09126794338226318,
"step": 777
},
{
"epoch": 0.9511002444987775,
"grad_norm": 0.562587320804596,
"learning_rate": 1.0786505312098975e-06,
"loss": 0.05866596847772598,
"step": 778
},
{
"epoch": 0.9523227383863081,
"grad_norm": 1.017979383468628,
"learning_rate": 1.0767360470431157e-06,
"loss": 0.15067419409751892,
"step": 779
},
{
"epoch": 0.9535452322738386,
"grad_norm": 0.5539723038673401,
"learning_rate": 1.0748212799115193e-06,
"loss": 0.0397900827229023,
"step": 780
},
{
"epoch": 0.9547677261613692,
"grad_norm": 0.5367475152015686,
"learning_rate": 1.0729062368758277e-06,
"loss": 0.058430396020412445,
"step": 781
},
{
"epoch": 0.9559902200488998,
"grad_norm": 0.9124805331230164,
"learning_rate": 1.0709909249977787e-06,
"loss": 0.13230904936790466,
"step": 782
},
{
"epoch": 0.9572127139364304,
"grad_norm": 0.5297884941101074,
"learning_rate": 1.0690753513401004e-06,
"loss": 0.045228056609630585,
"step": 783
},
{
"epoch": 0.9584352078239609,
"grad_norm": 1.81132972240448,
"learning_rate": 1.0671595229664865e-06,
"loss": 0.14689397811889648,
"step": 784
},
{
"epoch": 0.9596577017114915,
"grad_norm": 0.5824591517448425,
"learning_rate": 1.0652434469415703e-06,
"loss": 0.06070874258875847,
"step": 785
},
{
"epoch": 0.960880195599022,
"grad_norm": 1.5784008502960205,
"learning_rate": 1.0633271303308985e-06,
"loss": 0.20181557536125183,
"step": 786
},
{
"epoch": 0.9621026894865525,
"grad_norm": 0.5768479108810425,
"learning_rate": 1.0614105802009043e-06,
"loss": 0.03892522305250168,
"step": 787
},
{
"epoch": 0.9633251833740831,
"grad_norm": 0.4885495603084564,
"learning_rate": 1.0594938036188826e-06,
"loss": 0.04725014418363571,
"step": 788
},
{
"epoch": 0.9645476772616137,
"grad_norm": 1.272979736328125,
"learning_rate": 1.0575768076529625e-06,
"loss": 0.2130482792854309,
"step": 789
},
{
"epoch": 0.9657701711491442,
"grad_norm": 0.554162859916687,
"learning_rate": 1.055659599372083e-06,
"loss": 0.03410697728395462,
"step": 790
},
{
"epoch": 0.9669926650366748,
"grad_norm": 0.8635366559028625,
"learning_rate": 1.053742185845966e-06,
"loss": 0.10027478635311127,
"step": 791
},
{
"epoch": 0.9682151589242054,
"grad_norm": 0.603411078453064,
"learning_rate": 1.0518245741450894e-06,
"loss": 0.06285232305526733,
"step": 792
},
{
"epoch": 0.969437652811736,
"grad_norm": 0.539638102054596,
"learning_rate": 1.0499067713406622e-06,
"loss": 0.05545904487371445,
"step": 793
},
{
"epoch": 0.9706601466992665,
"grad_norm": 0.614665687084198,
"learning_rate": 1.0479887845045988e-06,
"loss": 0.060743607580661774,
"step": 794
},
{
"epoch": 0.9718826405867971,
"grad_norm": 1.2900402545928955,
"learning_rate": 1.046070620709492e-06,
"loss": 0.18782326579093933,
"step": 795
},
{
"epoch": 0.9731051344743277,
"grad_norm": 1.0307062864303589,
"learning_rate": 1.0441522870285858e-06,
"loss": 0.0633808895945549,
"step": 796
},
{
"epoch": 0.9743276283618582,
"grad_norm": 0.6567124724388123,
"learning_rate": 1.042233790535752e-06,
"loss": 0.0744098499417305,
"step": 797
},
{
"epoch": 0.9755501222493888,
"grad_norm": 0.3361664116382599,
"learning_rate": 1.040315138305463e-06,
"loss": 0.03247498720884323,
"step": 798
},
{
"epoch": 0.9767726161369193,
"grad_norm": 0.6204875111579895,
"learning_rate": 1.0383963374127645e-06,
"loss": 0.07268201559782028,
"step": 799
},
{
"epoch": 0.9779951100244498,
"grad_norm": 0.3996511697769165,
"learning_rate": 1.0364773949332508e-06,
"loss": 0.034229643642902374,
"step": 800
},
{
"epoch": 0.9792176039119804,
"grad_norm": 0.5463794469833374,
"learning_rate": 1.0345583179430385e-06,
"loss": 0.04507928341627121,
"step": 801
},
{
"epoch": 0.980440097799511,
"grad_norm": 1.8346048593521118,
"learning_rate": 1.03263911351874e-06,
"loss": 0.12306741625070572,
"step": 802
},
{
"epoch": 0.9816625916870416,
"grad_norm": 1.6830644607543945,
"learning_rate": 1.0307197887374374e-06,
"loss": 0.1579514443874359,
"step": 803
},
{
"epoch": 0.9828850855745721,
"grad_norm": 0.8887269496917725,
"learning_rate": 1.0288003506766574e-06,
"loss": 0.10700113326311111,
"step": 804
},
{
"epoch": 0.9841075794621027,
"grad_norm": 0.7811926603317261,
"learning_rate": 1.0268808064143436e-06,
"loss": 0.09710525721311569,
"step": 805
},
{
"epoch": 0.9853300733496333,
"grad_norm": 0.5428810715675354,
"learning_rate": 1.0249611630288313e-06,
"loss": 0.04861386865377426,
"step": 806
},
{
"epoch": 0.9865525672371638,
"grad_norm": 0.7606347799301147,
"learning_rate": 1.0230414275988218e-06,
"loss": 0.07365000247955322,
"step": 807
},
{
"epoch": 0.9877750611246944,
"grad_norm": 0.7765824198722839,
"learning_rate": 1.0211216072033557e-06,
"loss": 0.10606049001216888,
"step": 808
},
{
"epoch": 0.988997555012225,
"grad_norm": 0.7922176122665405,
"learning_rate": 1.0192017089217861e-06,
"loss": 0.029449842870235443,
"step": 809
},
{
"epoch": 0.9902200488997555,
"grad_norm": 0.4635094702243805,
"learning_rate": 1.0172817398337546e-06,
"loss": 0.05085951089859009,
"step": 810
},
{
"epoch": 0.991442542787286,
"grad_norm": 0.8666695356369019,
"learning_rate": 1.0153617070191632e-06,
"loss": 0.08809876441955566,
"step": 811
},
{
"epoch": 0.9926650366748166,
"grad_norm": 0.7485467195510864,
"learning_rate": 1.0134416175581483e-06,
"loss": 0.0813947319984436,
"step": 812
},
{
"epoch": 0.9938875305623472,
"grad_norm": 0.5183290839195251,
"learning_rate": 1.0115214785310566e-06,
"loss": 0.04236570745706558,
"step": 813
},
{
"epoch": 0.9951100244498777,
"grad_norm": 0.9470935463905334,
"learning_rate": 1.0096012970184164e-06,
"loss": 0.0926065593957901,
"step": 814
},
{
"epoch": 0.9963325183374083,
"grad_norm": 1.4170160293579102,
"learning_rate": 1.0076810801009137e-06,
"loss": 0.11892987042665482,
"step": 815
},
{
"epoch": 0.9975550122249389,
"grad_norm": 0.8529789447784424,
"learning_rate": 1.0057608348593634e-06,
"loss": 0.13535147905349731,
"step": 816
},
{
"epoch": 0.9987775061124694,
"grad_norm": 0.43919849395751953,
"learning_rate": 1.0038405683746867e-06,
"loss": 0.03170529752969742,
"step": 817
},
{
"epoch": 1.0,
"grad_norm": 2.122828483581543,
"learning_rate": 1.0019202877278818e-06,
"loss": 0.1556968241930008,
"step": 818
},
{
"epoch": 1.0012224938875305,
"grad_norm": 1.1944435834884644,
"learning_rate": 1e-06,
"loss": 0.17662441730499268,
"step": 819
},
{
"epoch": 1.0024449877750612,
"grad_norm": 0.8182088732719421,
"learning_rate": 9.980797122721183e-07,
"loss": 0.1100330576300621,
"step": 820
},
{
"epoch": 1.0036674816625917,
"grad_norm": 1.4344838857650757,
"learning_rate": 9.961594316253132e-07,
"loss": 0.15558874607086182,
"step": 821
},
{
"epoch": 1.0048899755501222,
"grad_norm": 0.45090049505233765,
"learning_rate": 9.942391651406365e-07,
"loss": 0.02767450362443924,
"step": 822
},
{
"epoch": 1.0061124694376529,
"grad_norm": 0.6195530295372009,
"learning_rate": 9.923189198990866e-07,
"loss": 0.09205645322799683,
"step": 823
},
{
"epoch": 1.0073349633251834,
"grad_norm": 0.49658119678497314,
"learning_rate": 9.903987029815833e-07,
"loss": 0.0548279732465744,
"step": 824
},
{
"epoch": 1.0085574572127138,
"grad_norm": 0.3632658123970032,
"learning_rate": 9.884785214689433e-07,
"loss": 0.02266687899827957,
"step": 825
},
{
"epoch": 1.0097799511002445,
"grad_norm": 0.731834888458252,
"learning_rate": 9.865583824418518e-07,
"loss": 0.08152560889720917,
"step": 826
},
{
"epoch": 1.011002444987775,
"grad_norm": 0.6704753637313843,
"learning_rate": 9.84638292980837e-07,
"loss": 0.05641212686896324,
"step": 827
},
{
"epoch": 1.0122249388753057,
"grad_norm": 1.1629999876022339,
"learning_rate": 9.827182601662456e-07,
"loss": 0.12735047936439514,
"step": 828
},
{
"epoch": 1.0134474327628362,
"grad_norm": 0.6338582038879395,
"learning_rate": 9.80798291078214e-07,
"loss": 0.06392369419336319,
"step": 829
},
{
"epoch": 1.0146699266503667,
"grad_norm": 0.6299936175346375,
"learning_rate": 9.788783927966445e-07,
"loss": 0.06995239853858948,
"step": 830
},
{
"epoch": 1.0158924205378974,
"grad_norm": 0.6832887530326843,
"learning_rate": 9.76958572401178e-07,
"loss": 0.08369270712137222,
"step": 831
},
{
"epoch": 1.017114914425428,
"grad_norm": 0.5458358526229858,
"learning_rate": 9.750388369711684e-07,
"loss": 0.061671994626522064,
"step": 832
},
{
"epoch": 1.0183374083129584,
"grad_norm": 0.418218195438385,
"learning_rate": 9.731191935856565e-07,
"loss": 0.0252443328499794,
"step": 833
},
{
"epoch": 1.019559902200489,
"grad_norm": 0.6674004793167114,
"learning_rate": 9.711996493233426e-07,
"loss": 0.06086137890815735,
"step": 834
},
{
"epoch": 1.0207823960880196,
"grad_norm": 1.34614098072052,
"learning_rate": 9.692802112625623e-07,
"loss": 0.09148378670215607,
"step": 835
},
{
"epoch": 1.02200488997555,
"grad_norm": 0.4576461613178253,
"learning_rate": 9.6736088648126e-07,
"loss": 0.04705827683210373,
"step": 836
},
{
"epoch": 1.0232273838630808,
"grad_norm": 0.7733202576637268,
"learning_rate": 9.654416820569616e-07,
"loss": 0.05014306306838989,
"step": 837
},
{
"epoch": 1.0244498777506112,
"grad_norm": 0.5602098107337952,
"learning_rate": 9.63522605066749e-07,
"loss": 0.04882945120334625,
"step": 838
},
{
"epoch": 1.0256723716381417,
"grad_norm": 0.9765346646308899,
"learning_rate": 9.616036625872356e-07,
"loss": 0.13497163355350494,
"step": 839
},
{
"epoch": 1.0268948655256724,
"grad_norm": 0.5365645289421082,
"learning_rate": 9.596848616945373e-07,
"loss": 0.03578435257077217,
"step": 840
},
{
"epoch": 1.028117359413203,
"grad_norm": 0.9197410345077515,
"learning_rate": 9.577662094642478e-07,
"loss": 0.059567756950855255,
"step": 841
},
{
"epoch": 1.0293398533007334,
"grad_norm": 0.5968348383903503,
"learning_rate": 9.558477129714143e-07,
"loss": 0.036009710282087326,
"step": 842
},
{
"epoch": 1.0305623471882641,
"grad_norm": 0.3897453546524048,
"learning_rate": 9.539293792905082e-07,
"loss": 0.018843837082386017,
"step": 843
},
{
"epoch": 1.0317848410757946,
"grad_norm": 0.6754714846611023,
"learning_rate": 9.520112154954009e-07,
"loss": 0.1061909943819046,
"step": 844
},
{
"epoch": 1.033007334963325,
"grad_norm": 0.4596664309501648,
"learning_rate": 9.500932286593378e-07,
"loss": 0.04471080005168915,
"step": 845
},
{
"epoch": 1.0342298288508558,
"grad_norm": 0.622141420841217,
"learning_rate": 9.481754258549108e-07,
"loss": 0.03550569713115692,
"step": 846
},
{
"epoch": 1.0354523227383863,
"grad_norm": 0.31733208894729614,
"learning_rate": 9.46257814154034e-07,
"loss": 0.021436432376503944,
"step": 847
},
{
"epoch": 1.0366748166259168,
"grad_norm": 0.7751713991165161,
"learning_rate": 9.44340400627917e-07,
"loss": 0.08037877082824707,
"step": 848
},
{
"epoch": 1.0378973105134475,
"grad_norm": 0.5616546273231506,
"learning_rate": 9.424231923470376e-07,
"loss": 0.03870062530040741,
"step": 849
},
{
"epoch": 1.039119804400978,
"grad_norm": 0.5779324769973755,
"learning_rate": 9.405061963811176e-07,
"loss": 0.049199171364307404,
"step": 850
},
{
"epoch": 1.0403422982885087,
"grad_norm": 0.5662124752998352,
"learning_rate": 9.385894197990956e-07,
"loss": 0.053786374628543854,
"step": 851
},
{
"epoch": 1.0415647921760391,
"grad_norm": 0.6696890592575073,
"learning_rate": 9.366728696691013e-07,
"loss": 0.06372671574354172,
"step": 852
},
{
"epoch": 1.0427872860635696,
"grad_norm": 1.2137998342514038,
"learning_rate": 9.347565530584297e-07,
"loss": 0.10669227689504623,
"step": 853
},
{
"epoch": 1.0440097799511003,
"grad_norm": 0.5666466951370239,
"learning_rate": 9.328404770335137e-07,
"loss": 0.07631044089794159,
"step": 854
},
{
"epoch": 1.0452322738386308,
"grad_norm": 0.577532947063446,
"learning_rate": 9.309246486598999e-07,
"loss": 0.07483428716659546,
"step": 855
},
{
"epoch": 1.0464547677261613,
"grad_norm": 0.9704628586769104,
"learning_rate": 9.290090750022212e-07,
"loss": 0.08475008606910706,
"step": 856
},
{
"epoch": 1.047677261613692,
"grad_norm": 0.6938967108726501,
"learning_rate": 9.270937631241722e-07,
"loss": 0.08225710690021515,
"step": 857
},
{
"epoch": 1.0488997555012225,
"grad_norm": 1.9143768548965454,
"learning_rate": 9.251787200884808e-07,
"loss": 0.06054507941007614,
"step": 858
},
{
"epoch": 1.050122249388753,
"grad_norm": 0.5811392068862915,
"learning_rate": 9.232639529568842e-07,
"loss": 0.07900646328926086,
"step": 859
},
{
"epoch": 1.0513447432762837,
"grad_norm": 0.49745267629623413,
"learning_rate": 9.213494687901026e-07,
"loss": 0.036775365471839905,
"step": 860
},
{
"epoch": 1.0525672371638142,
"grad_norm": 0.8511385321617126,
"learning_rate": 9.194352746478122e-07,
"loss": 0.08441373705863953,
"step": 861
},
{
"epoch": 1.0537897310513447,
"grad_norm": 0.3716083765029907,
"learning_rate": 9.175213775886195e-07,
"loss": 0.024708781391382217,
"step": 862
},
{
"epoch": 1.0550122249388754,
"grad_norm": 0.8983330130577087,
"learning_rate": 9.156077846700367e-07,
"loss": 0.09173930436372757,
"step": 863
},
{
"epoch": 1.0562347188264058,
"grad_norm": 0.7008864879608154,
"learning_rate": 9.136945029484526e-07,
"loss": 0.08458152413368225,
"step": 864
},
{
"epoch": 1.0574572127139363,
"grad_norm": 0.5859730839729309,
"learning_rate": 9.117815394791106e-07,
"loss": 0.05946192145347595,
"step": 865
},
{
"epoch": 1.058679706601467,
"grad_norm": 0.5610266327857971,
"learning_rate": 9.098689013160791e-07,
"loss": 0.04758323356509209,
"step": 866
},
{
"epoch": 1.0599022004889975,
"grad_norm": 0.582216739654541,
"learning_rate": 9.079565955122269e-07,
"loss": 0.058348044753074646,
"step": 867
},
{
"epoch": 1.061124694376528,
"grad_norm": 1.4693472385406494,
"learning_rate": 9.060446291191983e-07,
"loss": 0.08752306550741196,
"step": 868
},
{
"epoch": 1.0623471882640587,
"grad_norm": 0.3929154872894287,
"learning_rate": 9.041330091873851e-07,
"loss": 0.03629240766167641,
"step": 869
},
{
"epoch": 1.0635696821515892,
"grad_norm": 0.8240336775779724,
"learning_rate": 9.022217427659013e-07,
"loss": 0.09561021625995636,
"step": 870
},
{
"epoch": 1.06479217603912,
"grad_norm": 0.7420818209648132,
"learning_rate": 9.003108369025586e-07,
"loss": 0.09909369051456451,
"step": 871
},
{
"epoch": 1.0660146699266504,
"grad_norm": 0.7111018896102905,
"learning_rate": 8.984002986438374e-07,
"loss": 0.09067660570144653,
"step": 872
},
{
"epoch": 1.0672371638141809,
"grad_norm": 0.7000823616981506,
"learning_rate": 8.964901350348647e-07,
"loss": 0.07394730299711227,
"step": 873
},
{
"epoch": 1.0684596577017116,
"grad_norm": 2.0795962810516357,
"learning_rate": 8.94580353119384e-07,
"loss": 0.14223839342594147,
"step": 874
},
{
"epoch": 1.069682151589242,
"grad_norm": 0.6106052994728088,
"learning_rate": 8.926709599397318e-07,
"loss": 0.05609330162405968,
"step": 875
},
{
"epoch": 1.0709046454767726,
"grad_norm": 0.8647387027740479,
"learning_rate": 8.907619625368121e-07,
"loss": 0.1379162073135376,
"step": 876
},
{
"epoch": 1.0721271393643033,
"grad_norm": 0.92774897813797,
"learning_rate": 8.888533679500687e-07,
"loss": 0.1191943883895874,
"step": 877
},
{
"epoch": 1.0733496332518337,
"grad_norm": 0.4201011061668396,
"learning_rate": 8.869451832174592e-07,
"loss": 0.043148741126060486,
"step": 878
},
{
"epoch": 1.0745721271393642,
"grad_norm": 1.1211622953414917,
"learning_rate": 8.850374153754321e-07,
"loss": 0.22709359228610992,
"step": 879
},
{
"epoch": 1.075794621026895,
"grad_norm": 0.569976806640625,
"learning_rate": 8.831300714588964e-07,
"loss": 0.04401343688368797,
"step": 880
},
{
"epoch": 1.0770171149144254,
"grad_norm": 0.7433338165283203,
"learning_rate": 8.812231585011986e-07,
"loss": 0.08383987098932266,
"step": 881
},
{
"epoch": 1.078239608801956,
"grad_norm": 0.5324569940567017,
"learning_rate": 8.793166835340968e-07,
"loss": 0.047075316309928894,
"step": 882
},
{
"epoch": 1.0794621026894866,
"grad_norm": 0.5127405524253845,
"learning_rate": 8.77410653587733e-07,
"loss": 0.05578341335058212,
"step": 883
},
{
"epoch": 1.080684596577017,
"grad_norm": 0.6022531986236572,
"learning_rate": 8.755050756906084e-07,
"loss": 0.059749752283096313,
"step": 884
},
{
"epoch": 1.0819070904645476,
"grad_norm": 0.6054802536964417,
"learning_rate": 8.735999568695579e-07,
"loss": 0.10008956491947174,
"step": 885
},
{
"epoch": 1.0831295843520783,
"grad_norm": 0.32729411125183105,
"learning_rate": 8.716953041497228e-07,
"loss": 0.01694210059940815,
"step": 886
},
{
"epoch": 1.0843520782396088,
"grad_norm": 0.8257930278778076,
"learning_rate": 8.697911245545255e-07,
"loss": 0.08347071707248688,
"step": 887
},
{
"epoch": 1.0855745721271393,
"grad_norm": 0.8610199093818665,
"learning_rate": 8.67887425105645e-07,
"loss": 0.10332295298576355,
"step": 888
},
{
"epoch": 1.08679706601467,
"grad_norm": 0.6741382479667664,
"learning_rate": 8.659842128229886e-07,
"loss": 0.0553562194108963,
"step": 889
},
{
"epoch": 1.0880195599022005,
"grad_norm": 0.4833920896053314,
"learning_rate": 8.640814947246671e-07,
"loss": 0.030828863382339478,
"step": 890
},
{
"epoch": 1.089242053789731,
"grad_norm": 1.0182445049285889,
"learning_rate": 8.6217927782697e-07,
"loss": 0.09715455770492554,
"step": 891
},
{
"epoch": 1.0904645476772616,
"grad_norm": 0.9427770376205444,
"learning_rate": 8.602775691443373e-07,
"loss": 0.12197822332382202,
"step": 892
},
{
"epoch": 1.0916870415647921,
"grad_norm": 0.5352261066436768,
"learning_rate": 8.583763756893366e-07,
"loss": 0.035015448927879333,
"step": 893
},
{
"epoch": 1.0929095354523228,
"grad_norm": 0.8558014035224915,
"learning_rate": 8.564757044726339e-07,
"loss": 0.12189626693725586,
"step": 894
},
{
"epoch": 1.0941320293398533,
"grad_norm": 0.7817391753196716,
"learning_rate": 8.545755625029703e-07,
"loss": 0.08008717000484467,
"step": 895
},
{
"epoch": 1.0953545232273838,
"grad_norm": 0.6227569580078125,
"learning_rate": 8.526759567871356e-07,
"loss": 0.06369563937187195,
"step": 896
},
{
"epoch": 1.0965770171149145,
"grad_norm": 1.8583862781524658,
"learning_rate": 8.507768943299414e-07,
"loss": 0.21010063588619232,
"step": 897
},
{
"epoch": 1.097799511002445,
"grad_norm": 0.5088163614273071,
"learning_rate": 8.48878382134196e-07,
"loss": 0.045932866632938385,
"step": 898
},
{
"epoch": 1.0990220048899755,
"grad_norm": 0.747974693775177,
"learning_rate": 8.4698042720068e-07,
"loss": 0.1034088060259819,
"step": 899
},
{
"epoch": 1.1002444987775062,
"grad_norm": 0.5552902221679688,
"learning_rate": 8.450830365281177e-07,
"loss": 0.04663268104195595,
"step": 900
},
{
"epoch": 1.1014669926650367,
"grad_norm": 0.5343089699745178,
"learning_rate": 8.431862171131522e-07,
"loss": 0.0565609335899353,
"step": 901
},
{
"epoch": 1.1026894865525672,
"grad_norm": 1.7603362798690796,
"learning_rate": 8.412899759503221e-07,
"loss": 0.17238205671310425,
"step": 902
},
{
"epoch": 1.1039119804400979,
"grad_norm": 0.49976295232772827,
"learning_rate": 8.393943200320321e-07,
"loss": 0.03350896015763283,
"step": 903
},
{
"epoch": 1.1051344743276283,
"grad_norm": 0.5339570641517639,
"learning_rate": 8.374992563485288e-07,
"loss": 0.05165301263332367,
"step": 904
},
{
"epoch": 1.1063569682151588,
"grad_norm": 0.4499541223049164,
"learning_rate": 8.356047918878761e-07,
"loss": 0.05879964679479599,
"step": 905
},
{
"epoch": 1.1075794621026895,
"grad_norm": 0.4396633803844452,
"learning_rate": 8.337109336359274e-07,
"loss": 0.028295112773776054,
"step": 906
},
{
"epoch": 1.10880195599022,
"grad_norm": 0.4972207844257355,
"learning_rate": 8.318176885763001e-07,
"loss": 0.036725904792547226,
"step": 907
},
{
"epoch": 1.1100244498777505,
"grad_norm": 0.44628992676734924,
"learning_rate": 8.299250636903521e-07,
"loss": 0.06183379143476486,
"step": 908
},
{
"epoch": 1.1112469437652812,
"grad_norm": 0.6201710104942322,
"learning_rate": 8.280330659571531e-07,
"loss": 0.042233578860759735,
"step": 909
},
{
"epoch": 1.1124694376528117,
"grad_norm": 1.3206669092178345,
"learning_rate": 8.261417023534603e-07,
"loss": 0.13557545840740204,
"step": 910
},
{
"epoch": 1.1136919315403424,
"grad_norm": 0.7862096428871155,
"learning_rate": 8.242509798536934e-07,
"loss": 0.05970001593232155,
"step": 911
},
{
"epoch": 1.114914425427873,
"grad_norm": 0.6002734899520874,
"learning_rate": 8.223609054299066e-07,
"loss": 0.05253787338733673,
"step": 912
},
{
"epoch": 1.1161369193154034,
"grad_norm": 0.6898171901702881,
"learning_rate": 8.204714860517661e-07,
"loss": 0.05980631336569786,
"step": 913
},
{
"epoch": 1.117359413202934,
"grad_norm": 0.43978118896484375,
"learning_rate": 8.185827286865212e-07,
"loss": 0.01971302181482315,
"step": 914
},
{
"epoch": 1.1185819070904646,
"grad_norm": 0.593339204788208,
"learning_rate": 8.166946402989799e-07,
"loss": 0.0640174150466919,
"step": 915
},
{
"epoch": 1.119804400977995,
"grad_norm": 0.5551378726959229,
"learning_rate": 8.148072278514848e-07,
"loss": 0.037118107080459595,
"step": 916
},
{
"epoch": 1.1210268948655258,
"grad_norm": 0.41860201954841614,
"learning_rate": 8.129204983038845e-07,
"loss": 0.018294628709554672,
"step": 917
},
{
"epoch": 1.1222493887530562,
"grad_norm": 0.41209959983825684,
"learning_rate": 8.110344586135103e-07,
"loss": 0.02978842705488205,
"step": 918
},
{
"epoch": 1.1234718826405867,
"grad_norm": 1.3424336910247803,
"learning_rate": 8.091491157351493e-07,
"loss": 0.12522253394126892,
"step": 919
},
{
"epoch": 1.1246943765281174,
"grad_norm": 0.6225645542144775,
"learning_rate": 8.072644766210189e-07,
"loss": 0.07480103522539139,
"step": 920
},
{
"epoch": 1.125916870415648,
"grad_norm": 0.6825823783874512,
"learning_rate": 8.053805482207418e-07,
"loss": 0.09257323294878006,
"step": 921
},
{
"epoch": 1.1271393643031784,
"grad_norm": 0.8395339846611023,
"learning_rate": 8.034973374813195e-07,
"loss": 0.07987058162689209,
"step": 922
},
{
"epoch": 1.128361858190709,
"grad_norm": 0.48649898171424866,
"learning_rate": 8.016148513471076e-07,
"loss": 0.04196672514081001,
"step": 923
},
{
"epoch": 1.1295843520782396,
"grad_norm": 1.558870792388916,
"learning_rate": 7.997330967597892e-07,
"loss": 0.11281789839267731,
"step": 924
},
{
"epoch": 1.13080684596577,
"grad_norm": 0.5265726447105408,
"learning_rate": 7.978520806583502e-07,
"loss": 0.0458921380341053,
"step": 925
},
{
"epoch": 1.1320293398533008,
"grad_norm": 0.6330893039703369,
"learning_rate": 7.959718099790533e-07,
"loss": 0.07182339578866959,
"step": 926
},
{
"epoch": 1.1332518337408313,
"grad_norm": 0.5050659775733948,
"learning_rate": 7.940922916554121e-07,
"loss": 0.019945954903960228,
"step": 927
},
{
"epoch": 1.1344743276283618,
"grad_norm": 0.7883423566818237,
"learning_rate": 7.922135326181662e-07,
"loss": 0.08532335609197617,
"step": 928
},
{
"epoch": 1.1356968215158925,
"grad_norm": 6.213228702545166,
"learning_rate": 7.903355397952556e-07,
"loss": 0.09858151525259018,
"step": 929
},
{
"epoch": 1.136919315403423,
"grad_norm": 0.6395642161369324,
"learning_rate": 7.884583201117937e-07,
"loss": 0.07978861033916473,
"step": 930
},
{
"epoch": 1.1381418092909534,
"grad_norm": 0.5896817445755005,
"learning_rate": 7.865818804900448e-07,
"loss": 0.07659559696912766,
"step": 931
},
{
"epoch": 1.1393643031784841,
"grad_norm": 0.5216361880302429,
"learning_rate": 7.84706227849395e-07,
"loss": 0.043631620705127716,
"step": 932
},
{
"epoch": 1.1405867970660146,
"grad_norm": 0.6127905249595642,
"learning_rate": 7.828313691063292e-07,
"loss": 0.050032343715429306,
"step": 933
},
{
"epoch": 1.141809290953545,
"grad_norm": 0.6666812896728516,
"learning_rate": 7.80957311174405e-07,
"loss": 0.06281676888465881,
"step": 934
},
{
"epoch": 1.1430317848410758,
"grad_norm": 0.7285805940628052,
"learning_rate": 7.79084060964226e-07,
"loss": 0.09277459979057312,
"step": 935
},
{
"epoch": 1.1442542787286063,
"grad_norm": 0.4884703755378723,
"learning_rate": 7.772116253834189e-07,
"loss": 0.03467327728867531,
"step": 936
},
{
"epoch": 1.145476772616137,
"grad_norm": 0.8311352133750916,
"learning_rate": 7.753400113366051e-07,
"loss": 0.087074413895607,
"step": 937
},
{
"epoch": 1.1466992665036675,
"grad_norm": 2.2175540924072266,
"learning_rate": 7.734692257253765e-07,
"loss": 0.17994043231010437,
"step": 938
},
{
"epoch": 1.147921760391198,
"grad_norm": 0.6339080929756165,
"learning_rate": 7.715992754482718e-07,
"loss": 0.07415372133255005,
"step": 939
},
{
"epoch": 1.1491442542787287,
"grad_norm": 0.8799629807472229,
"learning_rate": 7.697301674007477e-07,
"loss": 0.11817336827516556,
"step": 940
},
{
"epoch": 1.1503667481662592,
"grad_norm": 0.618196964263916,
"learning_rate": 7.678619084751553e-07,
"loss": 0.09984515607357025,
"step": 941
},
{
"epoch": 1.1515892420537897,
"grad_norm": 0.5356141328811646,
"learning_rate": 7.659945055607157e-07,
"loss": 0.04796317592263222,
"step": 942
},
{
"epoch": 1.1528117359413204,
"grad_norm": 0.7146682739257812,
"learning_rate": 7.641279655434928e-07,
"loss": 0.0712418481707573,
"step": 943
},
{
"epoch": 1.1540342298288508,
"grad_norm": 0.817133367061615,
"learning_rate": 7.622622953063676e-07,
"loss": 0.10241246968507767,
"step": 944
},
{
"epoch": 1.1552567237163813,
"grad_norm": 0.3623119294643402,
"learning_rate": 7.603975017290157e-07,
"loss": 0.039422377943992615,
"step": 945
},
{
"epoch": 1.156479217603912,
"grad_norm": 0.5818023681640625,
"learning_rate": 7.585335916878786e-07,
"loss": 0.05019117519259453,
"step": 946
},
{
"epoch": 1.1577017114914425,
"grad_norm": 1.0506552457809448,
"learning_rate": 7.566705720561398e-07,
"loss": 0.08910926431417465,
"step": 947
},
{
"epoch": 1.1589242053789732,
"grad_norm": 0.6765831112861633,
"learning_rate": 7.548084497037005e-07,
"loss": 0.06334209442138672,
"step": 948
},
{
"epoch": 1.1601466992665037,
"grad_norm": 0.6223803758621216,
"learning_rate": 7.529472314971522e-07,
"loss": 0.07700802385807037,
"step": 949
},
{
"epoch": 1.1613691931540342,
"grad_norm": 0.5353448390960693,
"learning_rate": 7.510869242997517e-07,
"loss": 0.05452726036310196,
"step": 950
},
{
"epoch": 1.162591687041565,
"grad_norm": 0.4640703797340393,
"learning_rate": 7.492275349713988e-07,
"loss": 0.02945774793624878,
"step": 951
},
{
"epoch": 1.1638141809290954,
"grad_norm": 0.6415523290634155,
"learning_rate": 7.473690703686058e-07,
"loss": 0.07028944790363312,
"step": 952
},
{
"epoch": 1.1650366748166259,
"grad_norm": 0.6928954720497131,
"learning_rate": 7.455115373444778e-07,
"loss": 0.08268164098262787,
"step": 953
},
{
"epoch": 1.1662591687041566,
"grad_norm": 1.3478461503982544,
"learning_rate": 7.436549427486825e-07,
"loss": 0.10552765429019928,
"step": 954
},
{
"epoch": 1.167481662591687,
"grad_norm": 0.5524037480354309,
"learning_rate": 7.417992934274278e-07,
"loss": 0.05783731862902641,
"step": 955
},
{
"epoch": 1.1687041564792175,
"grad_norm": 1.8956823348999023,
"learning_rate": 7.399445962234369e-07,
"loss": 0.20652121305465698,
"step": 956
},
{
"epoch": 1.1699266503667483,
"grad_norm": 0.8729202151298523,
"learning_rate": 7.380908579759206e-07,
"loss": 0.07003243267536163,
"step": 957
},
{
"epoch": 1.1711491442542787,
"grad_norm": 0.5419681072235107,
"learning_rate": 7.362380855205541e-07,
"loss": 0.04234430938959122,
"step": 958
},
{
"epoch": 1.1723716381418092,
"grad_norm": 0.8602526187896729,
"learning_rate": 7.34386285689452e-07,
"loss": 0.09982462972402573,
"step": 959
},
{
"epoch": 1.17359413202934,
"grad_norm": 0.5677328705787659,
"learning_rate": 7.325354653111413e-07,
"loss": 0.054829493165016174,
"step": 960
},
{
"epoch": 1.1748166259168704,
"grad_norm": 0.6051965951919556,
"learning_rate": 7.306856312105369e-07,
"loss": 0.07003351300954819,
"step": 961
},
{
"epoch": 1.176039119804401,
"grad_norm": 0.6444430351257324,
"learning_rate": 7.288367902089187e-07,
"loss": 0.049501605331897736,
"step": 962
},
{
"epoch": 1.1772616136919316,
"grad_norm": 0.41739708185195923,
"learning_rate": 7.269889491239029e-07,
"loss": 0.030942771583795547,
"step": 963
},
{
"epoch": 1.178484107579462,
"grad_norm": 0.47497791051864624,
"learning_rate": 7.251421147694183e-07,
"loss": 0.02877943217754364,
"step": 964
},
{
"epoch": 1.1797066014669926,
"grad_norm": 0.6866270899772644,
"learning_rate": 7.23296293955683e-07,
"loss": 0.07550613582134247,
"step": 965
},
{
"epoch": 1.1809290953545233,
"grad_norm": 0.8445516228675842,
"learning_rate": 7.214514934891763e-07,
"loss": 0.0790812149643898,
"step": 966
},
{
"epoch": 1.1821515892420538,
"grad_norm": 1.652129888534546,
"learning_rate": 7.196077201726146e-07,
"loss": 0.1525307297706604,
"step": 967
},
{
"epoch": 1.1833740831295843,
"grad_norm": 0.6826167106628418,
"learning_rate": 7.177649808049287e-07,
"loss": 0.08445430546998978,
"step": 968
},
{
"epoch": 1.184596577017115,
"grad_norm": 1.0335001945495605,
"learning_rate": 7.159232821812347e-07,
"loss": 0.1456667184829712,
"step": 969
},
{
"epoch": 1.1858190709046454,
"grad_norm": 0.8647975921630859,
"learning_rate": 7.140826310928114e-07,
"loss": 0.09577184915542603,
"step": 970
},
{
"epoch": 1.187041564792176,
"grad_norm": 0.6034070253372192,
"learning_rate": 7.12243034327076e-07,
"loss": 0.04485723376274109,
"step": 971
},
{
"epoch": 1.1882640586797066,
"grad_norm": 0.9478015899658203,
"learning_rate": 7.104044986675561e-07,
"loss": 0.09477706998586655,
"step": 972
},
{
"epoch": 1.1894865525672371,
"grad_norm": 0.7184820771217346,
"learning_rate": 7.085670308938673e-07,
"loss": 0.09178591519594193,
"step": 973
},
{
"epoch": 1.1907090464547676,
"grad_norm": 0.8856843113899231,
"learning_rate": 7.067306377816882e-07,
"loss": 0.13793283700942993,
"step": 974
},
{
"epoch": 1.1919315403422983,
"grad_norm": 0.845316469669342,
"learning_rate": 7.048953261027327e-07,
"loss": 0.09184427559375763,
"step": 975
},
{
"epoch": 1.1931540342298288,
"grad_norm": 2.1853232383728027,
"learning_rate": 7.030611026247289e-07,
"loss": 0.13195018470287323,
"step": 976
},
{
"epoch": 1.1943765281173595,
"grad_norm": 0.5053343772888184,
"learning_rate": 7.012279741113908e-07,
"loss": 0.06105492636561394,
"step": 977
},
{
"epoch": 1.19559902200489,
"grad_norm": 0.4571700692176819,
"learning_rate": 6.993959473223945e-07,
"loss": 0.030309367924928665,
"step": 978
},
{
"epoch": 1.1968215158924205,
"grad_norm": 0.6352927088737488,
"learning_rate": 6.975650290133554e-07,
"loss": 0.05217323452234268,
"step": 979
},
{
"epoch": 1.1980440097799512,
"grad_norm": 0.6180278062820435,
"learning_rate": 6.957352259357992e-07,
"loss": 0.04761992767453194,
"step": 980
},
{
"epoch": 1.1992665036674817,
"grad_norm": 0.8236117959022522,
"learning_rate": 6.939065448371398e-07,
"loss": 0.05372278392314911,
"step": 981
},
{
"epoch": 1.2004889975550121,
"grad_norm": 0.5007245540618896,
"learning_rate": 6.920789924606547e-07,
"loss": 0.05247459560632706,
"step": 982
},
{
"epoch": 1.2017114914425429,
"grad_norm": 1.515081524848938,
"learning_rate": 6.902525755454581e-07,
"loss": 0.13622437417507172,
"step": 983
},
{
"epoch": 1.2029339853300733,
"grad_norm": 0.8081589937210083,
"learning_rate": 6.88427300826478e-07,
"loss": 0.10199341922998428,
"step": 984
},
{
"epoch": 1.2041564792176038,
"grad_norm": 0.4114532172679901,
"learning_rate": 6.866031750344301e-07,
"loss": 0.03341814503073692,
"step": 985
},
{
"epoch": 1.2053789731051345,
"grad_norm": 0.5318084359169006,
"learning_rate": 6.847802048957931e-07,
"loss": 0.044970422983169556,
"step": 986
},
{
"epoch": 1.206601466992665,
"grad_norm": 0.5375232100486755,
"learning_rate": 6.82958397132785e-07,
"loss": 0.05409299582242966,
"step": 987
},
{
"epoch": 1.2078239608801957,
"grad_norm": 1.087988018989563,
"learning_rate": 6.811377584633371e-07,
"loss": 0.06359348446130753,
"step": 988
},
{
"epoch": 1.2090464547677262,
"grad_norm": 0.47567427158355713,
"learning_rate": 6.793182956010699e-07,
"loss": 0.036037515848875046,
"step": 989
},
{
"epoch": 1.2102689486552567,
"grad_norm": 0.6344244480133057,
"learning_rate": 6.775000152552676e-07,
"loss": 0.09375251084566116,
"step": 990
},
{
"epoch": 1.2114914425427874,
"grad_norm": 0.5078109502792358,
"learning_rate": 6.756829241308541e-07,
"loss": 0.032816044986248016,
"step": 991
},
{
"epoch": 1.2127139364303179,
"grad_norm": 2.121873617172241,
"learning_rate": 6.738670289283686e-07,
"loss": 0.18621137738227844,
"step": 992
},
{
"epoch": 1.2139364303178484,
"grad_norm": 0.9619345664978027,
"learning_rate": 6.720523363439392e-07,
"loss": 0.07466273009777069,
"step": 993
},
{
"epoch": 1.215158924205379,
"grad_norm": 0.6141215562820435,
"learning_rate": 6.702388530692605e-07,
"loss": 0.08134876191616058,
"step": 994
},
{
"epoch": 1.2163814180929096,
"grad_norm": 0.7350115180015564,
"learning_rate": 6.684265857915667e-07,
"loss": 0.07135564833879471,
"step": 995
},
{
"epoch": 1.21760391198044,
"grad_norm": 0.5875363945960999,
"learning_rate": 6.66615541193609e-07,
"loss": 0.05108661949634552,
"step": 996
},
{
"epoch": 1.2188264058679708,
"grad_norm": 0.946712851524353,
"learning_rate": 6.648057259536296e-07,
"loss": 0.15040233731269836,
"step": 997
},
{
"epoch": 1.2200488997555012,
"grad_norm": 0.6048445701599121,
"learning_rate": 6.629971467453368e-07,
"loss": 0.07135920226573944,
"step": 998
},
{
"epoch": 1.2212713936430317,
"grad_norm": 0.3589957654476166,
"learning_rate": 6.611898102378818e-07,
"loss": 0.024296937510371208,
"step": 999
},
{
"epoch": 1.2224938875305624,
"grad_norm": 0.8074060082435608,
"learning_rate": 6.593837230958332e-07,
"loss": 0.08226868510246277,
"step": 1000
},
{
"epoch": 1.223716381418093,
"grad_norm": 0.5763094425201416,
"learning_rate": 6.57578891979152e-07,
"loss": 0.05438359081745148,
"step": 1001
},
{
"epoch": 1.2249388753056234,
"grad_norm": 0.48690176010131836,
"learning_rate": 6.557753235431688e-07,
"loss": 0.03222460672259331,
"step": 1002
},
{
"epoch": 1.226161369193154,
"grad_norm": 0.524236261844635,
"learning_rate": 6.539730244385567e-07,
"loss": 0.03836928680539131,
"step": 1003
},
{
"epoch": 1.2273838630806846,
"grad_norm": 0.5367822051048279,
"learning_rate": 6.521720013113086e-07,
"loss": 0.06549528241157532,
"step": 1004
},
{
"epoch": 1.228606356968215,
"grad_norm": 0.5317434668540955,
"learning_rate": 6.503722608027128e-07,
"loss": 0.05378490686416626,
"step": 1005
},
{
"epoch": 1.2298288508557458,
"grad_norm": 0.7317752838134766,
"learning_rate": 6.485738095493271e-07,
"loss": 0.09424787759780884,
"step": 1006
},
{
"epoch": 1.2310513447432763,
"grad_norm": 0.6902745366096497,
"learning_rate": 6.467766541829555e-07,
"loss": 0.05784755200147629,
"step": 1007
},
{
"epoch": 1.2322738386308068,
"grad_norm": 0.6072686314582825,
"learning_rate": 6.44980801330624e-07,
"loss": 0.042070455849170685,
"step": 1008
},
{
"epoch": 1.2334963325183375,
"grad_norm": 0.4769473671913147,
"learning_rate": 6.431862576145544e-07,
"loss": 0.0426953062415123,
"step": 1009
},
{
"epoch": 1.234718826405868,
"grad_norm": 0.5268250107765198,
"learning_rate": 6.413930296521416e-07,
"loss": 0.050640836358070374,
"step": 1010
},
{
"epoch": 1.2359413202933984,
"grad_norm": 0.5046558380126953,
"learning_rate": 6.396011240559293e-07,
"loss": 0.03945199400186539,
"step": 1011
},
{
"epoch": 1.2371638141809291,
"grad_norm": 1.1124933958053589,
"learning_rate": 6.37810547433584e-07,
"loss": 0.10693266242742538,
"step": 1012
},
{
"epoch": 1.2383863080684596,
"grad_norm": 0.37918198108673096,
"learning_rate": 6.360213063878714e-07,
"loss": 0.016333535313606262,
"step": 1013
},
{
"epoch": 1.23960880195599,
"grad_norm": 0.9932835102081299,
"learning_rate": 6.34233407516634e-07,
"loss": 0.08856149762868881,
"step": 1014
},
{
"epoch": 1.2408312958435208,
"grad_norm": 0.5905684232711792,
"learning_rate": 6.324468574127626e-07,
"loss": 0.049201227724552155,
"step": 1015
},
{
"epoch": 1.2420537897310513,
"grad_norm": 0.6306707859039307,
"learning_rate": 6.306616626641767e-07,
"loss": 0.05397559702396393,
"step": 1016
},
{
"epoch": 1.243276283618582,
"grad_norm": 0.6195847392082214,
"learning_rate": 6.288778298537966e-07,
"loss": 0.05494469031691551,
"step": 1017
},
{
"epoch": 1.2444987775061125,
"grad_norm": 0.5696724653244019,
"learning_rate": 6.270953655595199e-07,
"loss": 0.07787225395441055,
"step": 1018
},
{
"epoch": 1.245721271393643,
"grad_norm": 0.6183673739433289,
"learning_rate": 6.253142763541995e-07,
"loss": 0.058855678886175156,
"step": 1019
},
{
"epoch": 1.2469437652811737,
"grad_norm": 0.5052947998046875,
"learning_rate": 6.235345688056166e-07,
"loss": 0.03834133222699165,
"step": 1020
}
],
"logging_steps": 1,
"max_steps": 1636,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 10,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 2.5292580723439488e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}