test_CPTlsft2 / last-checkpoint /trainer_state.json
Ba2han's picture
Training in progress, step 5166, checkpoint
9941920 verified
Raw
History Blame Contribute Delete
457 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.8012971389717034,
"eval_steps": 500,
"global_step": 5166,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0006973865439266349,
"grad_norm": 0.31864917278289795,
"learning_rate": 2.9069767441860464e-07,
"loss": 2.7418713569641113,
"step": 2
},
{
"epoch": 0.0013947730878532698,
"grad_norm": 0.33852168917655945,
"learning_rate": 8.72093023255814e-07,
"loss": 2.7723140716552734,
"step": 4
},
{
"epoch": 0.002092159631779905,
"grad_norm": 0.36493492126464844,
"learning_rate": 1.4534883720930234e-06,
"loss": 2.81181001663208,
"step": 6
},
{
"epoch": 0.0027895461757065395,
"grad_norm": 0.5011733770370483,
"learning_rate": 2.0348837209302328e-06,
"loss": 3.0660762786865234,
"step": 8
},
{
"epoch": 0.0034869327196331745,
"grad_norm": 0.308676540851593,
"learning_rate": 2.616279069767442e-06,
"loss": 2.829026699066162,
"step": 10
},
{
"epoch": 0.00418431926355981,
"grad_norm": 0.381355881690979,
"learning_rate": 3.1976744186046516e-06,
"loss": 2.911247730255127,
"step": 12
},
{
"epoch": 0.004881705807486445,
"grad_norm": 0.3091634213924408,
"learning_rate": 3.7790697674418603e-06,
"loss": 2.5624561309814453,
"step": 14
},
{
"epoch": 0.005579092351413079,
"grad_norm": 0.39268556237220764,
"learning_rate": 4.36046511627907e-06,
"loss": 2.9684319496154785,
"step": 16
},
{
"epoch": 0.006276478895339714,
"grad_norm": 0.2759358584880829,
"learning_rate": 4.941860465116279e-06,
"loss": 2.742588520050049,
"step": 18
},
{
"epoch": 0.006973865439266349,
"grad_norm": 0.27893197536468506,
"learning_rate": 5.523255813953489e-06,
"loss": 2.7324178218841553,
"step": 20
},
{
"epoch": 0.007671251983192984,
"grad_norm": 0.382383793592453,
"learning_rate": 6.1046511627906975e-06,
"loss": 2.864603042602539,
"step": 22
},
{
"epoch": 0.00836863852711962,
"grad_norm": 0.48626744747161865,
"learning_rate": 6.686046511627907e-06,
"loss": 3.0625786781311035,
"step": 24
},
{
"epoch": 0.009066025071046254,
"grad_norm": 0.3646681010723114,
"learning_rate": 7.267441860465117e-06,
"loss": 2.5928220748901367,
"step": 26
},
{
"epoch": 0.00976341161497289,
"grad_norm": 0.3723832666873932,
"learning_rate": 7.848837209302325e-06,
"loss": 2.937344789505005,
"step": 28
},
{
"epoch": 0.010460798158899524,
"grad_norm": 0.3601192235946655,
"learning_rate": 8.430232558139535e-06,
"loss": 2.651121139526367,
"step": 30
},
{
"epoch": 0.011158184702826158,
"grad_norm": 0.34927698969841003,
"learning_rate": 9.011627906976745e-06,
"loss": 2.7281885147094727,
"step": 32
},
{
"epoch": 0.011855571246752794,
"grad_norm": 0.26186636090278625,
"learning_rate": 9.593023255813954e-06,
"loss": 2.7544922828674316,
"step": 34
},
{
"epoch": 0.012552957790679428,
"grad_norm": 0.25888198614120483,
"learning_rate": 1.0174418604651164e-05,
"loss": 2.7551393508911133,
"step": 36
},
{
"epoch": 0.013250344334606064,
"grad_norm": 0.4327206015586853,
"learning_rate": 1.0755813953488372e-05,
"loss": 2.984186887741089,
"step": 38
},
{
"epoch": 0.013947730878532698,
"grad_norm": 0.28884637355804443,
"learning_rate": 1.1337209302325581e-05,
"loss": 2.596083879470825,
"step": 40
},
{
"epoch": 0.014645117422459334,
"grad_norm": 0.5529292225837708,
"learning_rate": 1.1918604651162791e-05,
"loss": 2.845648765563965,
"step": 42
},
{
"epoch": 0.015342503966385968,
"grad_norm": 0.21446593105793,
"learning_rate": 1.25e-05,
"loss": 2.681359052658081,
"step": 44
},
{
"epoch": 0.016039890510312602,
"grad_norm": 0.36436358094215393,
"learning_rate": 1.308139534883721e-05,
"loss": 2.7329678535461426,
"step": 46
},
{
"epoch": 0.01673727705423924,
"grad_norm": 0.4239608347415924,
"learning_rate": 1.3662790697674418e-05,
"loss": 2.9083914756774902,
"step": 48
},
{
"epoch": 0.017434663598165874,
"grad_norm": 0.4936756491661072,
"learning_rate": 1.424418604651163e-05,
"loss": 2.846252679824829,
"step": 50
},
{
"epoch": 0.018132050142092508,
"grad_norm": 0.4550386965274811,
"learning_rate": 1.4825581395348839e-05,
"loss": 2.7459042072296143,
"step": 52
},
{
"epoch": 0.018829436686019142,
"grad_norm": 0.32297810912132263,
"learning_rate": 1.5406976744186045e-05,
"loss": 3.0400166511535645,
"step": 54
},
{
"epoch": 0.01952682322994578,
"grad_norm": 0.6890983581542969,
"learning_rate": 1.5988372093023257e-05,
"loss": 3.166825532913208,
"step": 56
},
{
"epoch": 0.020224209773872414,
"grad_norm": 0.38901615142822266,
"learning_rate": 1.6569767441860464e-05,
"loss": 2.93516206741333,
"step": 58
},
{
"epoch": 0.020921596317799048,
"grad_norm": 0.38375547528266907,
"learning_rate": 1.7151162790697676e-05,
"loss": 2.6337456703186035,
"step": 60
},
{
"epoch": 0.021618982861725682,
"grad_norm": 0.552964985370636,
"learning_rate": 1.7732558139534887e-05,
"loss": 2.9862496852874756,
"step": 62
},
{
"epoch": 0.022316369405652316,
"grad_norm": 0.27494218945503235,
"learning_rate": 1.831395348837209e-05,
"loss": 2.729008674621582,
"step": 64
},
{
"epoch": 0.023013755949578954,
"grad_norm": 0.3909671902656555,
"learning_rate": 1.8895348837209303e-05,
"loss": 2.952936887741089,
"step": 66
},
{
"epoch": 0.023711142493505588,
"grad_norm": 0.3981049358844757,
"learning_rate": 1.9476744186046514e-05,
"loss": 2.9518542289733887,
"step": 68
},
{
"epoch": 0.024408529037432222,
"grad_norm": 0.2676134705543518,
"learning_rate": 2.0058139534883722e-05,
"loss": 2.7456436157226562,
"step": 70
},
{
"epoch": 0.025105915581358856,
"grad_norm": 0.5440641641616821,
"learning_rate": 2.0639534883720933e-05,
"loss": 2.9664361476898193,
"step": 72
},
{
"epoch": 0.025803302125285494,
"grad_norm": 0.3179803490638733,
"learning_rate": 2.1220930232558138e-05,
"loss": 2.7757673263549805,
"step": 74
},
{
"epoch": 0.026500688669212128,
"grad_norm": 0.22457678616046906,
"learning_rate": 2.180232558139535e-05,
"loss": 2.6265127658843994,
"step": 76
},
{
"epoch": 0.027198075213138762,
"grad_norm": 0.5075873136520386,
"learning_rate": 2.238372093023256e-05,
"loss": 2.8070034980773926,
"step": 78
},
{
"epoch": 0.027895461757065396,
"grad_norm": 0.29639291763305664,
"learning_rate": 2.296511627906977e-05,
"loss": 2.81837797164917,
"step": 80
},
{
"epoch": 0.028592848300992034,
"grad_norm": 0.42572078108787537,
"learning_rate": 2.354651162790698e-05,
"loss": 2.9185187816619873,
"step": 82
},
{
"epoch": 0.029290234844918668,
"grad_norm": 0.23258890211582184,
"learning_rate": 2.4127906976744188e-05,
"loss": 2.612539529800415,
"step": 84
},
{
"epoch": 0.029987621388845302,
"grad_norm": 0.2520723342895508,
"learning_rate": 2.4709302325581396e-05,
"loss": 2.7379989624023438,
"step": 86
},
{
"epoch": 0.030685007932771936,
"grad_norm": 0.27651524543762207,
"learning_rate": 2.5290697674418607e-05,
"loss": 2.7794711589813232,
"step": 88
},
{
"epoch": 0.03138239447669857,
"grad_norm": 0.22654417157173157,
"learning_rate": 2.5872093023255818e-05,
"loss": 2.619795799255371,
"step": 90
},
{
"epoch": 0.032079781020625205,
"grad_norm": 0.30205756425857544,
"learning_rate": 2.6453488372093026e-05,
"loss": 2.730988025665283,
"step": 92
},
{
"epoch": 0.03277716756455184,
"grad_norm": 0.2712193727493286,
"learning_rate": 2.703488372093023e-05,
"loss": 2.7551698684692383,
"step": 94
},
{
"epoch": 0.03347455410847848,
"grad_norm": 0.22152888774871826,
"learning_rate": 2.7616279069767442e-05,
"loss": 2.7695436477661133,
"step": 96
},
{
"epoch": 0.034171940652405114,
"grad_norm": 0.2610715627670288,
"learning_rate": 2.8197674418604653e-05,
"loss": 2.480830192565918,
"step": 98
},
{
"epoch": 0.03486932719633175,
"grad_norm": 0.26491448283195496,
"learning_rate": 2.8779069767441864e-05,
"loss": 2.644742488861084,
"step": 100
},
{
"epoch": 0.03556671374025838,
"grad_norm": 0.242788165807724,
"learning_rate": 2.9360465116279072e-05,
"loss": 2.6238479614257812,
"step": 102
},
{
"epoch": 0.036264100284185016,
"grad_norm": 0.24330241978168488,
"learning_rate": 2.9941860465116277e-05,
"loss": 2.779409408569336,
"step": 104
},
{
"epoch": 0.03696148682811165,
"grad_norm": 0.25451141595840454,
"learning_rate": 3.052325581395349e-05,
"loss": 2.8484294414520264,
"step": 106
},
{
"epoch": 0.037658873372038285,
"grad_norm": 0.3185073435306549,
"learning_rate": 3.11046511627907e-05,
"loss": 2.9070987701416016,
"step": 108
},
{
"epoch": 0.03835625991596492,
"grad_norm": 0.2701241970062256,
"learning_rate": 3.168604651162791e-05,
"loss": 2.9335508346557617,
"step": 110
},
{
"epoch": 0.03905364645989156,
"grad_norm": 0.259211927652359,
"learning_rate": 3.226744186046512e-05,
"loss": 2.845470428466797,
"step": 112
},
{
"epoch": 0.039751033003818194,
"grad_norm": 0.3809325098991394,
"learning_rate": 3.284883720930233e-05,
"loss": 3.0224204063415527,
"step": 114
},
{
"epoch": 0.04044841954774483,
"grad_norm": 0.28319111466407776,
"learning_rate": 3.343023255813954e-05,
"loss": 2.734405040740967,
"step": 116
},
{
"epoch": 0.04114580609167146,
"grad_norm": 0.2592206299304962,
"learning_rate": 3.401162790697674e-05,
"loss": 2.7508089542388916,
"step": 118
},
{
"epoch": 0.041843192635598096,
"grad_norm": 0.26692506670951843,
"learning_rate": 3.4593023255813954e-05,
"loss": 2.6235299110412598,
"step": 120
},
{
"epoch": 0.04254057917952473,
"grad_norm": 0.26486581563949585,
"learning_rate": 3.5174418604651165e-05,
"loss": 2.8470211029052734,
"step": 122
},
{
"epoch": 0.043237965723451365,
"grad_norm": 0.2748745083808899,
"learning_rate": 3.5755813953488376e-05,
"loss": 2.683603286743164,
"step": 124
},
{
"epoch": 0.043935352267378,
"grad_norm": 0.3697609007358551,
"learning_rate": 3.633720930232558e-05,
"loss": 2.86184024810791,
"step": 126
},
{
"epoch": 0.04463273881130463,
"grad_norm": 0.21504926681518555,
"learning_rate": 3.691860465116279e-05,
"loss": 2.5714964866638184,
"step": 128
},
{
"epoch": 0.045330125355231274,
"grad_norm": 0.24733126163482666,
"learning_rate": 3.7500000000000003e-05,
"loss": 2.819786787033081,
"step": 130
},
{
"epoch": 0.04602751189915791,
"grad_norm": 0.2333776205778122,
"learning_rate": 3.8081395348837215e-05,
"loss": 2.8762454986572266,
"step": 132
},
{
"epoch": 0.04672489844308454,
"grad_norm": 0.24650025367736816,
"learning_rate": 3.866279069767442e-05,
"loss": 2.869750499725342,
"step": 134
},
{
"epoch": 0.047422284987011176,
"grad_norm": 0.2731642723083496,
"learning_rate": 3.924418604651163e-05,
"loss": 2.934140682220459,
"step": 136
},
{
"epoch": 0.04811967153093781,
"grad_norm": 0.2511177062988281,
"learning_rate": 3.9825581395348835e-05,
"loss": 2.6802401542663574,
"step": 138
},
{
"epoch": 0.048817058074864444,
"grad_norm": 0.2267242968082428,
"learning_rate": 4.0406976744186046e-05,
"loss": 2.632960557937622,
"step": 140
},
{
"epoch": 0.04951444461879108,
"grad_norm": 0.18804524838924408,
"learning_rate": 4.098837209302326e-05,
"loss": 2.548309087753296,
"step": 142
},
{
"epoch": 0.05021183116271771,
"grad_norm": 0.21063442528247833,
"learning_rate": 4.156976744186047e-05,
"loss": 2.745614528656006,
"step": 144
},
{
"epoch": 0.05090921770664435,
"grad_norm": 0.2846011817455292,
"learning_rate": 4.215116279069768e-05,
"loss": 2.6707184314727783,
"step": 146
},
{
"epoch": 0.05160660425057099,
"grad_norm": 0.23666012287139893,
"learning_rate": 4.2732558139534885e-05,
"loss": 2.8555541038513184,
"step": 148
},
{
"epoch": 0.05230399079449762,
"grad_norm": 0.2252434641122818,
"learning_rate": 4.3313953488372096e-05,
"loss": 2.808776378631592,
"step": 150
},
{
"epoch": 0.053001377338424256,
"grad_norm": 0.18965467810630798,
"learning_rate": 4.389534883720931e-05,
"loss": 2.7739152908325195,
"step": 152
},
{
"epoch": 0.05369876388235089,
"grad_norm": 0.20163702964782715,
"learning_rate": 4.447674418604651e-05,
"loss": 2.467848777770996,
"step": 154
},
{
"epoch": 0.054396150426277524,
"grad_norm": 0.19562198221683502,
"learning_rate": 4.505813953488372e-05,
"loss": 2.67034912109375,
"step": 156
},
{
"epoch": 0.05509353697020416,
"grad_norm": 0.22114789485931396,
"learning_rate": 4.563953488372093e-05,
"loss": 2.5959391593933105,
"step": 158
},
{
"epoch": 0.05579092351413079,
"grad_norm": 0.23299212753772736,
"learning_rate": 4.622093023255814e-05,
"loss": 2.6960949897766113,
"step": 160
},
{
"epoch": 0.05648831005805743,
"grad_norm": 0.21247857809066772,
"learning_rate": 4.680232558139535e-05,
"loss": 2.654393196105957,
"step": 162
},
{
"epoch": 0.05718569660198407,
"grad_norm": 0.243447944521904,
"learning_rate": 4.738372093023256e-05,
"loss": 2.6438827514648438,
"step": 164
},
{
"epoch": 0.0578830831459107,
"grad_norm": 0.2817672789096832,
"learning_rate": 4.796511627906977e-05,
"loss": 2.6078686714172363,
"step": 166
},
{
"epoch": 0.058580469689837336,
"grad_norm": 0.41085195541381836,
"learning_rate": 4.854651162790698e-05,
"loss": 2.7752044200897217,
"step": 168
},
{
"epoch": 0.05927785623376397,
"grad_norm": 0.2486799955368042,
"learning_rate": 4.912790697674419e-05,
"loss": 2.811298370361328,
"step": 170
},
{
"epoch": 0.059975242777690604,
"grad_norm": 0.3200376331806183,
"learning_rate": 4.97093023255814e-05,
"loss": 2.853541612625122,
"step": 172
},
{
"epoch": 0.06067262932161724,
"grad_norm": 0.2841547429561615,
"learning_rate": 5e-05,
"loss": 2.6713409423828125,
"step": 174
},
{
"epoch": 0.06137001586554387,
"grad_norm": 0.18580801784992218,
"learning_rate": 5e-05,
"loss": 2.590787172317505,
"step": 176
},
{
"epoch": 0.06206740240947051,
"grad_norm": 0.2215770035982132,
"learning_rate": 5e-05,
"loss": 2.7266483306884766,
"step": 178
},
{
"epoch": 0.06276478895339714,
"grad_norm": 0.3527434170246124,
"learning_rate": 5e-05,
"loss": 2.7592153549194336,
"step": 180
},
{
"epoch": 0.06346217549732378,
"grad_norm": 0.23568178713321686,
"learning_rate": 5e-05,
"loss": 2.753912925720215,
"step": 182
},
{
"epoch": 0.06415956204125041,
"grad_norm": 0.27784112095832825,
"learning_rate": 5e-05,
"loss": 2.7343976497650146,
"step": 184
},
{
"epoch": 0.06485694858517704,
"grad_norm": 0.24607530236244202,
"learning_rate": 5e-05,
"loss": 2.3938684463500977,
"step": 186
},
{
"epoch": 0.06555433512910368,
"grad_norm": 0.2644323408603668,
"learning_rate": 5e-05,
"loss": 2.7110233306884766,
"step": 188
},
{
"epoch": 0.06625172167303033,
"grad_norm": 0.20555633306503296,
"learning_rate": 5e-05,
"loss": 2.5770535469055176,
"step": 190
},
{
"epoch": 0.06694910821695696,
"grad_norm": 0.25855734944343567,
"learning_rate": 5e-05,
"loss": 2.740647554397583,
"step": 192
},
{
"epoch": 0.0676464947608836,
"grad_norm": 0.1986989974975586,
"learning_rate": 5e-05,
"loss": 2.5619754791259766,
"step": 194
},
{
"epoch": 0.06834388130481023,
"grad_norm": 0.15838097035884857,
"learning_rate": 5e-05,
"loss": 2.6336307525634766,
"step": 196
},
{
"epoch": 0.06904126784873686,
"grad_norm": 0.23912972211837769,
"learning_rate": 5e-05,
"loss": 2.550100326538086,
"step": 198
},
{
"epoch": 0.0697386543926635,
"grad_norm": 0.23773403465747833,
"learning_rate": 5e-05,
"loss": 2.5678303241729736,
"step": 200
},
{
"epoch": 0.07043604093659013,
"grad_norm": 0.47105443477630615,
"learning_rate": 5e-05,
"loss": 2.7644424438476562,
"step": 202
},
{
"epoch": 0.07113342748051676,
"grad_norm": 0.21215203404426575,
"learning_rate": 5e-05,
"loss": 2.636762857437134,
"step": 204
},
{
"epoch": 0.0718308140244434,
"grad_norm": 0.2100985199213028,
"learning_rate": 5e-05,
"loss": 2.597527027130127,
"step": 206
},
{
"epoch": 0.07252820056837003,
"grad_norm": 0.18938426673412323,
"learning_rate": 5e-05,
"loss": 2.5953726768493652,
"step": 208
},
{
"epoch": 0.07322558711229667,
"grad_norm": 0.21786312758922577,
"learning_rate": 5e-05,
"loss": 2.5680484771728516,
"step": 210
},
{
"epoch": 0.0739229736562233,
"grad_norm": 0.1661650538444519,
"learning_rate": 5e-05,
"loss": 2.4869861602783203,
"step": 212
},
{
"epoch": 0.07462036020014993,
"grad_norm": 0.16630461812019348,
"learning_rate": 5e-05,
"loss": 2.3208518028259277,
"step": 214
},
{
"epoch": 0.07531774674407657,
"grad_norm": 0.2707706391811371,
"learning_rate": 5e-05,
"loss": 2.546509027481079,
"step": 216
},
{
"epoch": 0.0760151332880032,
"grad_norm": 0.1831883043050766,
"learning_rate": 5e-05,
"loss": 2.5652148723602295,
"step": 218
},
{
"epoch": 0.07671251983192984,
"grad_norm": 0.26978302001953125,
"learning_rate": 5e-05,
"loss": 2.8050994873046875,
"step": 220
},
{
"epoch": 0.07740990637585647,
"grad_norm": 0.2312367558479309,
"learning_rate": 5e-05,
"loss": 2.4646224975585938,
"step": 222
},
{
"epoch": 0.07810729291978312,
"grad_norm": 0.17823274433612823,
"learning_rate": 5e-05,
"loss": 2.317629814147949,
"step": 224
},
{
"epoch": 0.07880467946370975,
"grad_norm": 0.2108280509710312,
"learning_rate": 5e-05,
"loss": 2.501734733581543,
"step": 226
},
{
"epoch": 0.07950206600763639,
"grad_norm": 0.2048916518688202,
"learning_rate": 5e-05,
"loss": 2.677302360534668,
"step": 228
},
{
"epoch": 0.08019945255156302,
"grad_norm": 0.32073312997817993,
"learning_rate": 5e-05,
"loss": 2.731003999710083,
"step": 230
},
{
"epoch": 0.08089683909548966,
"grad_norm": 0.18938776850700378,
"learning_rate": 5e-05,
"loss": 2.5771291255950928,
"step": 232
},
{
"epoch": 0.08159422563941629,
"grad_norm": 0.2064923495054245,
"learning_rate": 5e-05,
"loss": 2.657177209854126,
"step": 234
},
{
"epoch": 0.08229161218334292,
"grad_norm": 0.27688175439834595,
"learning_rate": 5e-05,
"loss": 2.460599422454834,
"step": 236
},
{
"epoch": 0.08298899872726956,
"grad_norm": 0.16016334295272827,
"learning_rate": 5e-05,
"loss": 2.3141236305236816,
"step": 238
},
{
"epoch": 0.08368638527119619,
"grad_norm": 0.17556940019130707,
"learning_rate": 5e-05,
"loss": 2.45051646232605,
"step": 240
},
{
"epoch": 0.08438377181512283,
"grad_norm": 0.22553177177906036,
"learning_rate": 5e-05,
"loss": 2.3532402515411377,
"step": 242
},
{
"epoch": 0.08508115835904946,
"grad_norm": 0.17820724844932556,
"learning_rate": 5e-05,
"loss": 2.5057754516601562,
"step": 244
},
{
"epoch": 0.0857785449029761,
"grad_norm": 0.18560370802879333,
"learning_rate": 5e-05,
"loss": 2.528789520263672,
"step": 246
},
{
"epoch": 0.08647593144690273,
"grad_norm": 0.15187770128250122,
"learning_rate": 5e-05,
"loss": 2.4031500816345215,
"step": 248
},
{
"epoch": 0.08717331799082936,
"grad_norm": 0.15653961896896362,
"learning_rate": 5e-05,
"loss": 2.504204511642456,
"step": 250
},
{
"epoch": 0.087870704534756,
"grad_norm": 0.16170385479927063,
"learning_rate": 5e-05,
"loss": 2.4689741134643555,
"step": 252
},
{
"epoch": 0.08856809107868263,
"grad_norm": 0.1795680671930313,
"learning_rate": 5e-05,
"loss": 2.513284921646118,
"step": 254
},
{
"epoch": 0.08926547762260927,
"grad_norm": 0.19295957684516907,
"learning_rate": 5e-05,
"loss": 2.615536689758301,
"step": 256
},
{
"epoch": 0.08996286416653591,
"grad_norm": 0.15200206637382507,
"learning_rate": 5e-05,
"loss": 2.289851665496826,
"step": 258
},
{
"epoch": 0.09066025071046255,
"grad_norm": 0.25096359848976135,
"learning_rate": 5e-05,
"loss": 2.6007142066955566,
"step": 260
},
{
"epoch": 0.09135763725438918,
"grad_norm": 0.18293362855911255,
"learning_rate": 5e-05,
"loss": 2.6895923614501953,
"step": 262
},
{
"epoch": 0.09205502379831582,
"grad_norm": 0.19729362428188324,
"learning_rate": 5e-05,
"loss": 2.628530263900757,
"step": 264
},
{
"epoch": 0.09275241034224245,
"grad_norm": 0.18690696358680725,
"learning_rate": 5e-05,
"loss": 2.5449976921081543,
"step": 266
},
{
"epoch": 0.09344979688616908,
"grad_norm": 0.15589185059070587,
"learning_rate": 5e-05,
"loss": 2.4761691093444824,
"step": 268
},
{
"epoch": 0.09414718343009572,
"grad_norm": 0.18014448881149292,
"learning_rate": 5e-05,
"loss": 2.541571617126465,
"step": 270
},
{
"epoch": 0.09484456997402235,
"grad_norm": 0.3075288236141205,
"learning_rate": 5e-05,
"loss": 2.4309425354003906,
"step": 272
},
{
"epoch": 0.09554195651794899,
"grad_norm": 0.2191179245710373,
"learning_rate": 5e-05,
"loss": 2.4452948570251465,
"step": 274
},
{
"epoch": 0.09623934306187562,
"grad_norm": 0.22002685070037842,
"learning_rate": 5e-05,
"loss": 2.439091205596924,
"step": 276
},
{
"epoch": 0.09693672960580225,
"grad_norm": 0.2415190041065216,
"learning_rate": 5e-05,
"loss": 2.459465980529785,
"step": 278
},
{
"epoch": 0.09763411614972889,
"grad_norm": 0.14990724623203278,
"learning_rate": 5e-05,
"loss": 2.4255549907684326,
"step": 280
},
{
"epoch": 0.09833150269365552,
"grad_norm": 0.19191303849220276,
"learning_rate": 5e-05,
"loss": 2.36759352684021,
"step": 282
},
{
"epoch": 0.09902888923758216,
"grad_norm": 0.18174131214618683,
"learning_rate": 5e-05,
"loss": 2.142644166946411,
"step": 284
},
{
"epoch": 0.09972627578150879,
"grad_norm": 0.29124879837036133,
"learning_rate": 5e-05,
"loss": 2.5742549896240234,
"step": 286
},
{
"epoch": 0.10042366232543543,
"grad_norm": 0.2132851630449295,
"learning_rate": 5e-05,
"loss": 2.541057586669922,
"step": 288
},
{
"epoch": 0.10112104886936206,
"grad_norm": 0.17444126307964325,
"learning_rate": 5e-05,
"loss": 2.5035133361816406,
"step": 290
},
{
"epoch": 0.1018184354132887,
"grad_norm": 0.21964430809020996,
"learning_rate": 5e-05,
"loss": 2.4692187309265137,
"step": 292
},
{
"epoch": 0.10251582195721534,
"grad_norm": 0.1695927530527115,
"learning_rate": 5e-05,
"loss": 2.4870972633361816,
"step": 294
},
{
"epoch": 0.10321320850114198,
"grad_norm": 0.1802135556936264,
"learning_rate": 5e-05,
"loss": 2.330629825592041,
"step": 296
},
{
"epoch": 0.10391059504506861,
"grad_norm": 0.20426394045352936,
"learning_rate": 5e-05,
"loss": 2.526278257369995,
"step": 298
},
{
"epoch": 0.10460798158899524,
"grad_norm": 0.24609476327896118,
"learning_rate": 5e-05,
"loss": 2.57684326171875,
"step": 300
},
{
"epoch": 0.10530536813292188,
"grad_norm": 0.24746456742286682,
"learning_rate": 5e-05,
"loss": 2.5122692584991455,
"step": 302
},
{
"epoch": 0.10600275467684851,
"grad_norm": 0.1913493573665619,
"learning_rate": 5e-05,
"loss": 2.1479685306549072,
"step": 304
},
{
"epoch": 0.10670014122077515,
"grad_norm": 0.1576538383960724,
"learning_rate": 5e-05,
"loss": 2.3806564807891846,
"step": 306
},
{
"epoch": 0.10739752776470178,
"grad_norm": 0.16610626876354218,
"learning_rate": 5e-05,
"loss": 2.5583462715148926,
"step": 308
},
{
"epoch": 0.10809491430862841,
"grad_norm": 0.16851180791854858,
"learning_rate": 5e-05,
"loss": 2.2103683948516846,
"step": 310
},
{
"epoch": 0.10879230085255505,
"grad_norm": 0.1957131028175354,
"learning_rate": 5e-05,
"loss": 2.457911968231201,
"step": 312
},
{
"epoch": 0.10948968739648168,
"grad_norm": 0.1593063920736313,
"learning_rate": 5e-05,
"loss": 2.4814860820770264,
"step": 314
},
{
"epoch": 0.11018707394040832,
"grad_norm": 0.22903546690940857,
"learning_rate": 5e-05,
"loss": 2.6023154258728027,
"step": 316
},
{
"epoch": 0.11088446048433495,
"grad_norm": 0.177312433719635,
"learning_rate": 5e-05,
"loss": 2.4045705795288086,
"step": 318
},
{
"epoch": 0.11158184702826159,
"grad_norm": 0.20729391276836395,
"learning_rate": 5e-05,
"loss": 2.299720048904419,
"step": 320
},
{
"epoch": 0.11227923357218822,
"grad_norm": 0.19588986039161682,
"learning_rate": 5e-05,
"loss": 2.3049607276916504,
"step": 322
},
{
"epoch": 0.11297662011611485,
"grad_norm": 0.166270911693573,
"learning_rate": 5e-05,
"loss": 2.4332315921783447,
"step": 324
},
{
"epoch": 0.11367400666004149,
"grad_norm": 0.15577951073646545,
"learning_rate": 5e-05,
"loss": 2.1308932304382324,
"step": 326
},
{
"epoch": 0.11437139320396814,
"grad_norm": 0.20020805299282074,
"learning_rate": 5e-05,
"loss": 2.4692726135253906,
"step": 328
},
{
"epoch": 0.11506877974789477,
"grad_norm": 0.16950419545173645,
"learning_rate": 5e-05,
"loss": 2.3831939697265625,
"step": 330
},
{
"epoch": 0.1157661662918214,
"grad_norm": 0.26503539085388184,
"learning_rate": 5e-05,
"loss": 2.479071617126465,
"step": 332
},
{
"epoch": 0.11646355283574804,
"grad_norm": 0.15860703587532043,
"learning_rate": 5e-05,
"loss": 2.2117795944213867,
"step": 334
},
{
"epoch": 0.11716093937967467,
"grad_norm": 0.1510804444551468,
"learning_rate": 5e-05,
"loss": 2.3502540588378906,
"step": 336
},
{
"epoch": 0.1178583259236013,
"grad_norm": 0.1815343201160431,
"learning_rate": 5e-05,
"loss": 2.080516815185547,
"step": 338
},
{
"epoch": 0.11855571246752794,
"grad_norm": 0.18012766540050507,
"learning_rate": 5e-05,
"loss": 2.4024839401245117,
"step": 340
},
{
"epoch": 0.11925309901145457,
"grad_norm": 0.2472798079252243,
"learning_rate": 5e-05,
"loss": 2.454730987548828,
"step": 342
},
{
"epoch": 0.11995048555538121,
"grad_norm": 0.18788838386535645,
"learning_rate": 5e-05,
"loss": 2.409149408340454,
"step": 344
},
{
"epoch": 0.12064787209930784,
"grad_norm": 0.16861069202423096,
"learning_rate": 5e-05,
"loss": 2.5427353382110596,
"step": 346
},
{
"epoch": 0.12134525864323448,
"grad_norm": 0.1705676168203354,
"learning_rate": 5e-05,
"loss": 2.322288990020752,
"step": 348
},
{
"epoch": 0.12204264518716111,
"grad_norm": 0.1752922534942627,
"learning_rate": 5e-05,
"loss": 2.363172769546509,
"step": 350
},
{
"epoch": 0.12274003173108775,
"grad_norm": 0.15499891340732574,
"learning_rate": 5e-05,
"loss": 2.448575973510742,
"step": 352
},
{
"epoch": 0.12343741827501438,
"grad_norm": 0.16332626342773438,
"learning_rate": 5e-05,
"loss": 2.51427960395813,
"step": 354
},
{
"epoch": 0.12413480481894101,
"grad_norm": 0.16457317769527435,
"learning_rate": 5e-05,
"loss": 2.299191474914551,
"step": 356
},
{
"epoch": 0.12483219136286765,
"grad_norm": 0.1974119395017624,
"learning_rate": 5e-05,
"loss": 2.3599853515625,
"step": 358
},
{
"epoch": 0.12552957790679428,
"grad_norm": 0.17325985431671143,
"learning_rate": 5e-05,
"loss": 2.3927571773529053,
"step": 360
},
{
"epoch": 0.12622696445072093,
"grad_norm": 0.2036568522453308,
"learning_rate": 5e-05,
"loss": 2.5239930152893066,
"step": 362
},
{
"epoch": 0.12692435099464755,
"grad_norm": 0.17068889737129211,
"learning_rate": 5e-05,
"loss": 2.2897727489471436,
"step": 364
},
{
"epoch": 0.1276217375385742,
"grad_norm": 0.1659744679927826,
"learning_rate": 5e-05,
"loss": 2.254037618637085,
"step": 366
},
{
"epoch": 0.12831912408250082,
"grad_norm": 0.19931660592556,
"learning_rate": 5e-05,
"loss": 2.3429343700408936,
"step": 368
},
{
"epoch": 0.12901651062642747,
"grad_norm": 0.19183991849422455,
"learning_rate": 5e-05,
"loss": 2.1451992988586426,
"step": 370
},
{
"epoch": 0.1297138971703541,
"grad_norm": 0.1866455227136612,
"learning_rate": 5e-05,
"loss": 2.4415817260742188,
"step": 372
},
{
"epoch": 0.13041128371428073,
"grad_norm": 0.15720690786838531,
"learning_rate": 5e-05,
"loss": 2.091867208480835,
"step": 374
},
{
"epoch": 0.13110867025820735,
"grad_norm": 0.185997873544693,
"learning_rate": 5e-05,
"loss": 2.3023877143859863,
"step": 376
},
{
"epoch": 0.131806056802134,
"grad_norm": 0.20472486317157745,
"learning_rate": 5e-05,
"loss": 2.394613027572632,
"step": 378
},
{
"epoch": 0.13250344334606065,
"grad_norm": 0.17751619219779968,
"learning_rate": 5e-05,
"loss": 2.2397282123565674,
"step": 380
},
{
"epoch": 0.13320082988998727,
"grad_norm": 0.23776665329933167,
"learning_rate": 5e-05,
"loss": 2.221808433532715,
"step": 382
},
{
"epoch": 0.13389821643391392,
"grad_norm": 0.2056867778301239,
"learning_rate": 5e-05,
"loss": 2.2870006561279297,
"step": 384
},
{
"epoch": 0.13459560297784054,
"grad_norm": 0.1593138575553894,
"learning_rate": 5e-05,
"loss": 2.419555187225342,
"step": 386
},
{
"epoch": 0.1352929895217672,
"grad_norm": 0.23647361993789673,
"learning_rate": 5e-05,
"loss": 2.2779324054718018,
"step": 388
},
{
"epoch": 0.1359903760656938,
"grad_norm": 0.3032076060771942,
"learning_rate": 5e-05,
"loss": 2.2887792587280273,
"step": 390
},
{
"epoch": 0.13668776260962046,
"grad_norm": 0.2255440354347229,
"learning_rate": 5e-05,
"loss": 2.287172794342041,
"step": 392
},
{
"epoch": 0.13738514915354708,
"grad_norm": 0.20240047574043274,
"learning_rate": 5e-05,
"loss": 2.4954910278320312,
"step": 394
},
{
"epoch": 0.13808253569747372,
"grad_norm": 0.15618708729743958,
"learning_rate": 5e-05,
"loss": 2.3185558319091797,
"step": 396
},
{
"epoch": 0.13877992224140034,
"grad_norm": 0.23974068462848663,
"learning_rate": 5e-05,
"loss": 2.3013062477111816,
"step": 398
},
{
"epoch": 0.139477308785327,
"grad_norm": 0.1627752184867859,
"learning_rate": 5e-05,
"loss": 2.240105628967285,
"step": 400
},
{
"epoch": 0.1401746953292536,
"grad_norm": 0.27172815799713135,
"learning_rate": 5e-05,
"loss": 2.3200438022613525,
"step": 402
},
{
"epoch": 0.14087208187318026,
"grad_norm": 0.26541343331336975,
"learning_rate": 5e-05,
"loss": 2.315690517425537,
"step": 404
},
{
"epoch": 0.14156946841710688,
"grad_norm": 0.17275846004486084,
"learning_rate": 5e-05,
"loss": 2.258578300476074,
"step": 406
},
{
"epoch": 0.14226685496103353,
"grad_norm": 0.18345215916633606,
"learning_rate": 5e-05,
"loss": 2.2334933280944824,
"step": 408
},
{
"epoch": 0.14296424150496015,
"grad_norm": 0.1952318549156189,
"learning_rate": 5e-05,
"loss": 2.368171453475952,
"step": 410
},
{
"epoch": 0.1436616280488868,
"grad_norm": 0.17698706686496735,
"learning_rate": 5e-05,
"loss": 2.5310282707214355,
"step": 412
},
{
"epoch": 0.14435901459281344,
"grad_norm": 0.18241949379444122,
"learning_rate": 5e-05,
"loss": 2.263460636138916,
"step": 414
},
{
"epoch": 0.14505640113674007,
"grad_norm": 0.1891452819108963,
"learning_rate": 5e-05,
"loss": 2.341625213623047,
"step": 416
},
{
"epoch": 0.1457537876806667,
"grad_norm": 0.13853508234024048,
"learning_rate": 5e-05,
"loss": 2.274247169494629,
"step": 418
},
{
"epoch": 0.14645117422459333,
"grad_norm": 0.25757747888565063,
"learning_rate": 5e-05,
"loss": 2.3942487239837646,
"step": 420
},
{
"epoch": 0.14714856076851998,
"grad_norm": 0.32313287258148193,
"learning_rate": 5e-05,
"loss": 2.181455135345459,
"step": 422
},
{
"epoch": 0.1478459473124466,
"grad_norm": 0.28559526801109314,
"learning_rate": 5e-05,
"loss": 2.2236557006835938,
"step": 424
},
{
"epoch": 0.14854333385637325,
"grad_norm": 0.24414172768592834,
"learning_rate": 5e-05,
"loss": 2.165065050125122,
"step": 426
},
{
"epoch": 0.14924072040029987,
"grad_norm": 0.22570635378360748,
"learning_rate": 5e-05,
"loss": 2.509464979171753,
"step": 428
},
{
"epoch": 0.14993810694422652,
"grad_norm": 0.18143275380134583,
"learning_rate": 5e-05,
"loss": 2.3109960556030273,
"step": 430
},
{
"epoch": 0.15063549348815314,
"grad_norm": 0.20212240517139435,
"learning_rate": 5e-05,
"loss": 2.2753820419311523,
"step": 432
},
{
"epoch": 0.1513328800320798,
"grad_norm": 0.2698517441749573,
"learning_rate": 5e-05,
"loss": 2.2871546745300293,
"step": 434
},
{
"epoch": 0.1520302665760064,
"grad_norm": 0.17344266176223755,
"learning_rate": 5e-05,
"loss": 2.178943634033203,
"step": 436
},
{
"epoch": 0.15272765311993305,
"grad_norm": 0.2650235891342163,
"learning_rate": 5e-05,
"loss": 2.276524782180786,
"step": 438
},
{
"epoch": 0.15342503966385967,
"grad_norm": 0.2071993052959442,
"learning_rate": 5e-05,
"loss": 2.3213233947753906,
"step": 440
},
{
"epoch": 0.15412242620778632,
"grad_norm": 0.1874510943889618,
"learning_rate": 5e-05,
"loss": 2.30794095993042,
"step": 442
},
{
"epoch": 0.15481981275171294,
"grad_norm": 0.20554737746715546,
"learning_rate": 5e-05,
"loss": 2.3562545776367188,
"step": 444
},
{
"epoch": 0.1555171992956396,
"grad_norm": 0.18512044847011566,
"learning_rate": 5e-05,
"loss": 2.169607639312744,
"step": 446
},
{
"epoch": 0.15621458583956624,
"grad_norm": 0.20362405478954315,
"learning_rate": 5e-05,
"loss": 2.1602494716644287,
"step": 448
},
{
"epoch": 0.15691197238349286,
"grad_norm": 0.2092098444700241,
"learning_rate": 5e-05,
"loss": 2.316654682159424,
"step": 450
},
{
"epoch": 0.1576093589274195,
"grad_norm": 0.24519668519496918,
"learning_rate": 5e-05,
"loss": 2.3567047119140625,
"step": 452
},
{
"epoch": 0.15830674547134613,
"grad_norm": 0.21312598884105682,
"learning_rate": 5e-05,
"loss": 2.227585792541504,
"step": 454
},
{
"epoch": 0.15900413201527278,
"grad_norm": 0.22957709431648254,
"learning_rate": 5e-05,
"loss": 2.2935221195220947,
"step": 456
},
{
"epoch": 0.1597015185591994,
"grad_norm": 0.18547071516513824,
"learning_rate": 5e-05,
"loss": 2.1790616512298584,
"step": 458
},
{
"epoch": 0.16039890510312604,
"grad_norm": 0.3164200186729431,
"learning_rate": 5e-05,
"loss": 2.3586549758911133,
"step": 460
},
{
"epoch": 0.16109629164705266,
"grad_norm": 0.15056589245796204,
"learning_rate": 5e-05,
"loss": 2.197247266769409,
"step": 462
},
{
"epoch": 0.1617936781909793,
"grad_norm": 0.1786457598209381,
"learning_rate": 5e-05,
"loss": 2.3103506565093994,
"step": 464
},
{
"epoch": 0.16249106473490593,
"grad_norm": 0.1755671203136444,
"learning_rate": 5e-05,
"loss": 2.2560946941375732,
"step": 466
},
{
"epoch": 0.16318845127883258,
"grad_norm": 0.2025860697031021,
"learning_rate": 5e-05,
"loss": 2.3827438354492188,
"step": 468
},
{
"epoch": 0.1638858378227592,
"grad_norm": 0.27311503887176514,
"learning_rate": 5e-05,
"loss": 2.2892584800720215,
"step": 470
},
{
"epoch": 0.16458322436668585,
"grad_norm": 0.1848417967557907,
"learning_rate": 5e-05,
"loss": 1.9864113330841064,
"step": 472
},
{
"epoch": 0.16528061091061247,
"grad_norm": 0.22180335223674774,
"learning_rate": 5e-05,
"loss": 2.1797947883605957,
"step": 474
},
{
"epoch": 0.16597799745453912,
"grad_norm": 0.17293785512447357,
"learning_rate": 5e-05,
"loss": 2.1941161155700684,
"step": 476
},
{
"epoch": 0.16667538399846574,
"grad_norm": 0.18438875675201416,
"learning_rate": 5e-05,
"loss": 2.2773759365081787,
"step": 478
},
{
"epoch": 0.16737277054239239,
"grad_norm": 0.23384010791778564,
"learning_rate": 5e-05,
"loss": 2.2826220989227295,
"step": 480
},
{
"epoch": 0.16807015708631903,
"grad_norm": 0.18001523613929749,
"learning_rate": 5e-05,
"loss": 2.1910438537597656,
"step": 482
},
{
"epoch": 0.16876754363024565,
"grad_norm": 0.17984063923358917,
"learning_rate": 5e-05,
"loss": 2.3001465797424316,
"step": 484
},
{
"epoch": 0.1694649301741723,
"grad_norm": 0.20125970244407654,
"learning_rate": 5e-05,
"loss": 2.288634777069092,
"step": 486
},
{
"epoch": 0.17016231671809892,
"grad_norm": 0.21922867000102997,
"learning_rate": 5e-05,
"loss": 2.0868029594421387,
"step": 488
},
{
"epoch": 0.17085970326202557,
"grad_norm": 0.1705346703529358,
"learning_rate": 5e-05,
"loss": 2.288616895675659,
"step": 490
},
{
"epoch": 0.1715570898059522,
"grad_norm": 0.21018339693546295,
"learning_rate": 5e-05,
"loss": 2.2519397735595703,
"step": 492
},
{
"epoch": 0.17225447634987884,
"grad_norm": 0.2216828316450119,
"learning_rate": 5e-05,
"loss": 2.0601437091827393,
"step": 494
},
{
"epoch": 0.17295186289380546,
"grad_norm": 0.18288859724998474,
"learning_rate": 5e-05,
"loss": 2.2289626598358154,
"step": 496
},
{
"epoch": 0.1736492494377321,
"grad_norm": 0.20881079137325287,
"learning_rate": 5e-05,
"loss": 2.1628737449645996,
"step": 498
},
{
"epoch": 0.17434663598165873,
"grad_norm": 0.2011629045009613,
"learning_rate": 5e-05,
"loss": 2.113814353942871,
"step": 500
},
{
"epoch": 0.17504402252558537,
"grad_norm": 0.20714020729064941,
"learning_rate": 5e-05,
"loss": 2.1722474098205566,
"step": 502
},
{
"epoch": 0.175741409069512,
"grad_norm": 0.18272532522678375,
"learning_rate": 5e-05,
"loss": 2.204862356185913,
"step": 504
},
{
"epoch": 0.17643879561343864,
"grad_norm": 0.29607588052749634,
"learning_rate": 5e-05,
"loss": 2.1915314197540283,
"step": 506
},
{
"epoch": 0.17713618215736526,
"grad_norm": 0.18939310312271118,
"learning_rate": 5e-05,
"loss": 2.3054800033569336,
"step": 508
},
{
"epoch": 0.1778335687012919,
"grad_norm": 0.1929270178079605,
"learning_rate": 5e-05,
"loss": 2.2519474029541016,
"step": 510
},
{
"epoch": 0.17853095524521853,
"grad_norm": 0.207675963640213,
"learning_rate": 5e-05,
"loss": 2.4564568996429443,
"step": 512
},
{
"epoch": 0.17922834178914518,
"grad_norm": 0.17932309210300446,
"learning_rate": 5e-05,
"loss": 2.211622953414917,
"step": 514
},
{
"epoch": 0.17992572833307183,
"grad_norm": 0.22701098024845123,
"learning_rate": 5e-05,
"loss": 2.292675256729126,
"step": 516
},
{
"epoch": 0.18062311487699845,
"grad_norm": 0.22169411182403564,
"learning_rate": 5e-05,
"loss": 2.3756208419799805,
"step": 518
},
{
"epoch": 0.1813205014209251,
"grad_norm": 0.2705889046192169,
"learning_rate": 5e-05,
"loss": 2.043933153152466,
"step": 520
},
{
"epoch": 0.18201788796485172,
"grad_norm": 0.21304632723331451,
"learning_rate": 5e-05,
"loss": 2.251720428466797,
"step": 522
},
{
"epoch": 0.18271527450877836,
"grad_norm": 0.2548201084136963,
"learning_rate": 5e-05,
"loss": 2.201152801513672,
"step": 524
},
{
"epoch": 0.18341266105270498,
"grad_norm": 0.21620337665081024,
"learning_rate": 5e-05,
"loss": 2.335510730743408,
"step": 526
},
{
"epoch": 0.18411004759663163,
"grad_norm": 0.20087069272994995,
"learning_rate": 5e-05,
"loss": 2.2359421253204346,
"step": 528
},
{
"epoch": 0.18480743414055825,
"grad_norm": 0.16715870797634125,
"learning_rate": 5e-05,
"loss": 2.2858028411865234,
"step": 530
},
{
"epoch": 0.1855048206844849,
"grad_norm": 0.18232391774654388,
"learning_rate": 5e-05,
"loss": 2.2304000854492188,
"step": 532
},
{
"epoch": 0.18620220722841152,
"grad_norm": 0.20358413457870483,
"learning_rate": 5e-05,
"loss": 2.268289566040039,
"step": 534
},
{
"epoch": 0.18689959377233817,
"grad_norm": 0.20649784803390503,
"learning_rate": 5e-05,
"loss": 2.1258363723754883,
"step": 536
},
{
"epoch": 0.1875969803162648,
"grad_norm": 0.22944477200508118,
"learning_rate": 5e-05,
"loss": 2.0966506004333496,
"step": 538
},
{
"epoch": 0.18829436686019144,
"grad_norm": 0.14751775562763214,
"learning_rate": 5e-05,
"loss": 2.4623894691467285,
"step": 540
},
{
"epoch": 0.18899175340411806,
"grad_norm": 0.16408704221248627,
"learning_rate": 5e-05,
"loss": 2.2561984062194824,
"step": 542
},
{
"epoch": 0.1896891399480447,
"grad_norm": 0.14852581918239594,
"learning_rate": 5e-05,
"loss": 2.139472007751465,
"step": 544
},
{
"epoch": 0.19038652649197133,
"grad_norm": 0.13257473707199097,
"learning_rate": 5e-05,
"loss": 2.087822675704956,
"step": 546
},
{
"epoch": 0.19108391303589797,
"grad_norm": 0.26599791646003723,
"learning_rate": 5e-05,
"loss": 2.2869253158569336,
"step": 548
},
{
"epoch": 0.1917812995798246,
"grad_norm": 0.23565323650836945,
"learning_rate": 5e-05,
"loss": 2.181511402130127,
"step": 550
},
{
"epoch": 0.19247868612375124,
"grad_norm": 0.1710377186536789,
"learning_rate": 5e-05,
"loss": 2.2027649879455566,
"step": 552
},
{
"epoch": 0.1931760726676779,
"grad_norm": 0.2599605321884155,
"learning_rate": 5e-05,
"loss": 2.0101318359375,
"step": 554
},
{
"epoch": 0.1938734592116045,
"grad_norm": 0.23988552391529083,
"learning_rate": 5e-05,
"loss": 2.2785067558288574,
"step": 556
},
{
"epoch": 0.19457084575553116,
"grad_norm": 0.19026997685432434,
"learning_rate": 5e-05,
"loss": 2.264369487762451,
"step": 558
},
{
"epoch": 0.19526823229945778,
"grad_norm": 0.21166720986366272,
"learning_rate": 5e-05,
"loss": 2.144094467163086,
"step": 560
},
{
"epoch": 0.19596561884338443,
"grad_norm": 0.17970100045204163,
"learning_rate": 5e-05,
"loss": 2.4488015174865723,
"step": 562
},
{
"epoch": 0.19666300538731105,
"grad_norm": 0.2090035080909729,
"learning_rate": 5e-05,
"loss": 2.2831969261169434,
"step": 564
},
{
"epoch": 0.1973603919312377,
"grad_norm": 0.13870304822921753,
"learning_rate": 5e-05,
"loss": 2.252763032913208,
"step": 566
},
{
"epoch": 0.19805777847516431,
"grad_norm": 0.1855902373790741,
"learning_rate": 5e-05,
"loss": 2.129446029663086,
"step": 568
},
{
"epoch": 0.19875516501909096,
"grad_norm": 0.29074016213417053,
"learning_rate": 5e-05,
"loss": 2.321843147277832,
"step": 570
},
{
"epoch": 0.19945255156301758,
"grad_norm": 0.2053501307964325,
"learning_rate": 5e-05,
"loss": 2.1515793800354004,
"step": 572
},
{
"epoch": 0.20014993810694423,
"grad_norm": 0.16916127502918243,
"learning_rate": 5e-05,
"loss": 2.0299763679504395,
"step": 574
},
{
"epoch": 0.20084732465087085,
"grad_norm": 0.18850381672382355,
"learning_rate": 5e-05,
"loss": 2.131117105484009,
"step": 576
},
{
"epoch": 0.2015447111947975,
"grad_norm": 0.27086082100868225,
"learning_rate": 5e-05,
"loss": 2.1912379264831543,
"step": 578
},
{
"epoch": 0.20224209773872412,
"grad_norm": 0.27545830607414246,
"learning_rate": 5e-05,
"loss": 2.19915771484375,
"step": 580
},
{
"epoch": 0.20293948428265077,
"grad_norm": 0.2752111256122589,
"learning_rate": 5e-05,
"loss": 2.267606258392334,
"step": 582
},
{
"epoch": 0.2036368708265774,
"grad_norm": 0.319011926651001,
"learning_rate": 5e-05,
"loss": 2.2916200160980225,
"step": 584
},
{
"epoch": 0.20433425737050404,
"grad_norm": 0.20431022346019745,
"learning_rate": 5e-05,
"loss": 2.2031617164611816,
"step": 586
},
{
"epoch": 0.20503164391443068,
"grad_norm": 0.2223842591047287,
"learning_rate": 5e-05,
"loss": 2.196582317352295,
"step": 588
},
{
"epoch": 0.2057290304583573,
"grad_norm": 0.3272736370563507,
"learning_rate": 5e-05,
"loss": 2.1547532081604004,
"step": 590
},
{
"epoch": 0.20642641700228395,
"grad_norm": 0.18153901398181915,
"learning_rate": 5e-05,
"loss": 2.2748570442199707,
"step": 592
},
{
"epoch": 0.20712380354621057,
"grad_norm": 0.20026224851608276,
"learning_rate": 5e-05,
"loss": 2.185814619064331,
"step": 594
},
{
"epoch": 0.20782119009013722,
"grad_norm": 0.2095404714345932,
"learning_rate": 5e-05,
"loss": 2.116464138031006,
"step": 596
},
{
"epoch": 0.20851857663406384,
"grad_norm": 0.2254505455493927,
"learning_rate": 5e-05,
"loss": 2.290566921234131,
"step": 598
},
{
"epoch": 0.2092159631779905,
"grad_norm": 0.22954094409942627,
"learning_rate": 5e-05,
"loss": 2.1632447242736816,
"step": 600
},
{
"epoch": 0.2099133497219171,
"grad_norm": 0.17814919352531433,
"learning_rate": 5e-05,
"loss": 2.1754729747772217,
"step": 602
},
{
"epoch": 0.21061073626584376,
"grad_norm": 0.31458622217178345,
"learning_rate": 5e-05,
"loss": 2.4768474102020264,
"step": 604
},
{
"epoch": 0.21130812280977038,
"grad_norm": 0.16490019857883453,
"learning_rate": 5e-05,
"loss": 2.1874232292175293,
"step": 606
},
{
"epoch": 0.21200550935369702,
"grad_norm": 0.2257475107908249,
"learning_rate": 5e-05,
"loss": 2.2535409927368164,
"step": 608
},
{
"epoch": 0.21270289589762365,
"grad_norm": 0.2660852074623108,
"learning_rate": 5e-05,
"loss": 2.061890125274658,
"step": 610
},
{
"epoch": 0.2134002824415503,
"grad_norm": 0.2156171202659607,
"learning_rate": 5e-05,
"loss": 2.2851383686065674,
"step": 612
},
{
"epoch": 0.2140976689854769,
"grad_norm": 0.17925716936588287,
"learning_rate": 5e-05,
"loss": 2.082984447479248,
"step": 614
},
{
"epoch": 0.21479505552940356,
"grad_norm": 0.3436058759689331,
"learning_rate": 5e-05,
"loss": 2.1233651638031006,
"step": 616
},
{
"epoch": 0.21549244207333018,
"grad_norm": 0.22150059044361115,
"learning_rate": 5e-05,
"loss": 2.023638963699341,
"step": 618
},
{
"epoch": 0.21618982861725683,
"grad_norm": 0.19689801335334778,
"learning_rate": 5e-05,
"loss": 2.082303762435913,
"step": 620
},
{
"epoch": 0.21688721516118348,
"grad_norm": 0.1721586138010025,
"learning_rate": 5e-05,
"loss": 2.1851935386657715,
"step": 622
},
{
"epoch": 0.2175846017051101,
"grad_norm": 0.21074773371219635,
"learning_rate": 5e-05,
"loss": 2.125735282897949,
"step": 624
},
{
"epoch": 0.21828198824903675,
"grad_norm": 0.2333507388830185,
"learning_rate": 5e-05,
"loss": 2.28065824508667,
"step": 626
},
{
"epoch": 0.21897937479296337,
"grad_norm": 0.20217208564281464,
"learning_rate": 5e-05,
"loss": 2.1805195808410645,
"step": 628
},
{
"epoch": 0.21967676133689001,
"grad_norm": 0.21829625964164734,
"learning_rate": 5e-05,
"loss": 1.9504680633544922,
"step": 630
},
{
"epoch": 0.22037414788081663,
"grad_norm": 0.19575423002243042,
"learning_rate": 5e-05,
"loss": 2.1475088596343994,
"step": 632
},
{
"epoch": 0.22107153442474328,
"grad_norm": 0.23382477462291718,
"learning_rate": 5e-05,
"loss": 2.1880345344543457,
"step": 634
},
{
"epoch": 0.2217689209686699,
"grad_norm": 0.2851170599460602,
"learning_rate": 5e-05,
"loss": 2.126394033432007,
"step": 636
},
{
"epoch": 0.22246630751259655,
"grad_norm": 0.2504940927028656,
"learning_rate": 5e-05,
"loss": 2.028315544128418,
"step": 638
},
{
"epoch": 0.22316369405652317,
"grad_norm": 0.1850201040506363,
"learning_rate": 5e-05,
"loss": 2.234729766845703,
"step": 640
},
{
"epoch": 0.22386108060044982,
"grad_norm": 0.15266118943691254,
"learning_rate": 5e-05,
"loss": 2.186972141265869,
"step": 642
},
{
"epoch": 0.22455846714437644,
"grad_norm": 0.2082221508026123,
"learning_rate": 5e-05,
"loss": 2.3294734954833984,
"step": 644
},
{
"epoch": 0.2252558536883031,
"grad_norm": 0.2343757152557373,
"learning_rate": 5e-05,
"loss": 2.3024473190307617,
"step": 646
},
{
"epoch": 0.2259532402322297,
"grad_norm": 0.27744945883750916,
"learning_rate": 5e-05,
"loss": 2.151794910430908,
"step": 648
},
{
"epoch": 0.22665062677615636,
"grad_norm": 0.19519643485546112,
"learning_rate": 5e-05,
"loss": 2.2723190784454346,
"step": 650
},
{
"epoch": 0.22734801332008298,
"grad_norm": 0.21738861501216888,
"learning_rate": 5e-05,
"loss": 2.1555187702178955,
"step": 652
},
{
"epoch": 0.22804539986400962,
"grad_norm": 0.2636173367500305,
"learning_rate": 5e-05,
"loss": 2.214590072631836,
"step": 654
},
{
"epoch": 0.22874278640793627,
"grad_norm": 0.21495908498764038,
"learning_rate": 5e-05,
"loss": 2.2480390071868896,
"step": 656
},
{
"epoch": 0.2294401729518629,
"grad_norm": 0.2296120971441269,
"learning_rate": 5e-05,
"loss": 2.1904211044311523,
"step": 658
},
{
"epoch": 0.23013755949578954,
"grad_norm": 0.30581578612327576,
"learning_rate": 5e-05,
"loss": 2.1074323654174805,
"step": 660
},
{
"epoch": 0.23083494603971616,
"grad_norm": 0.19764243066310883,
"learning_rate": 5e-05,
"loss": 2.223814010620117,
"step": 662
},
{
"epoch": 0.2315323325836428,
"grad_norm": 0.21603018045425415,
"learning_rate": 5e-05,
"loss": 2.140528440475464,
"step": 664
},
{
"epoch": 0.23222971912756943,
"grad_norm": 0.1800299435853958,
"learning_rate": 5e-05,
"loss": 2.1906323432922363,
"step": 666
},
{
"epoch": 0.23292710567149608,
"grad_norm": 0.24724136292934418,
"learning_rate": 5e-05,
"loss": 2.120924949645996,
"step": 668
},
{
"epoch": 0.2336244922154227,
"grad_norm": 0.2484823316335678,
"learning_rate": 5e-05,
"loss": 2.010047435760498,
"step": 670
},
{
"epoch": 0.23432187875934934,
"grad_norm": 0.19774702191352844,
"learning_rate": 5e-05,
"loss": 2.1234774589538574,
"step": 672
},
{
"epoch": 0.23501926530327596,
"grad_norm": 0.2553282678127289,
"learning_rate": 5e-05,
"loss": 2.1027908325195312,
"step": 674
},
{
"epoch": 0.2357166518472026,
"grad_norm": 0.20291560888290405,
"learning_rate": 5e-05,
"loss": 2.1387977600097656,
"step": 676
},
{
"epoch": 0.23641403839112923,
"grad_norm": 0.34688642621040344,
"learning_rate": 5e-05,
"loss": 2.1452786922454834,
"step": 678
},
{
"epoch": 0.23711142493505588,
"grad_norm": 0.2841176688671112,
"learning_rate": 5e-05,
"loss": 1.9979159832000732,
"step": 680
},
{
"epoch": 0.2378088114789825,
"grad_norm": 0.21012328565120697,
"learning_rate": 5e-05,
"loss": 2.271960735321045,
"step": 682
},
{
"epoch": 0.23850619802290915,
"grad_norm": 0.161057710647583,
"learning_rate": 5e-05,
"loss": 2.0357096195220947,
"step": 684
},
{
"epoch": 0.23920358456683577,
"grad_norm": 0.2356063425540924,
"learning_rate": 5e-05,
"loss": 2.1391124725341797,
"step": 686
},
{
"epoch": 0.23990097111076242,
"grad_norm": 0.21011163294315338,
"learning_rate": 5e-05,
"loss": 2.1928865909576416,
"step": 688
},
{
"epoch": 0.24059835765468907,
"grad_norm": 0.2434798926115036,
"learning_rate": 5e-05,
"loss": 2.017970323562622,
"step": 690
},
{
"epoch": 0.24129574419861569,
"grad_norm": 0.24937564134597778,
"learning_rate": 5e-05,
"loss": 2.058408260345459,
"step": 692
},
{
"epoch": 0.24199313074254233,
"grad_norm": 0.2534058094024658,
"learning_rate": 5e-05,
"loss": 2.002953052520752,
"step": 694
},
{
"epoch": 0.24269051728646895,
"grad_norm": 0.35094359517097473,
"learning_rate": 5e-05,
"loss": 2.2217190265655518,
"step": 696
},
{
"epoch": 0.2433879038303956,
"grad_norm": 0.19374975562095642,
"learning_rate": 5e-05,
"loss": 2.1322073936462402,
"step": 698
},
{
"epoch": 0.24408529037432222,
"grad_norm": 0.4570079445838928,
"learning_rate": 5e-05,
"loss": 2.106457233428955,
"step": 700
},
{
"epoch": 0.24478267691824887,
"grad_norm": 0.2247321903705597,
"learning_rate": 5e-05,
"loss": 2.0081517696380615,
"step": 702
},
{
"epoch": 0.2454800634621755,
"grad_norm": 0.21547703444957733,
"learning_rate": 5e-05,
"loss": 2.137990713119507,
"step": 704
},
{
"epoch": 0.24617745000610214,
"grad_norm": 0.253587007522583,
"learning_rate": 5e-05,
"loss": 2.056151866912842,
"step": 706
},
{
"epoch": 0.24687483655002876,
"grad_norm": 0.3514871895313263,
"learning_rate": 5e-05,
"loss": 2.067183494567871,
"step": 708
},
{
"epoch": 0.2475722230939554,
"grad_norm": 0.18376290798187256,
"learning_rate": 5e-05,
"loss": 2.1726388931274414,
"step": 710
},
{
"epoch": 0.24826960963788203,
"grad_norm": 0.2481544315814972,
"learning_rate": 5e-05,
"loss": 2.035372257232666,
"step": 712
},
{
"epoch": 0.24896699618180868,
"grad_norm": 0.23347629606723785,
"learning_rate": 5e-05,
"loss": 2.2434043884277344,
"step": 714
},
{
"epoch": 0.2496643827257353,
"grad_norm": 0.2897323668003082,
"learning_rate": 5e-05,
"loss": 2.0160436630249023,
"step": 716
},
{
"epoch": 0.2503617692696619,
"grad_norm": 0.24364791810512543,
"learning_rate": 5e-05,
"loss": 2.09956955909729,
"step": 718
},
{
"epoch": 0.25105915581358856,
"grad_norm": 0.24172277748584747,
"learning_rate": 5e-05,
"loss": 2.1436986923217773,
"step": 720
},
{
"epoch": 0.2517565423575152,
"grad_norm": 0.24258746206760406,
"learning_rate": 5e-05,
"loss": 2.0807886123657227,
"step": 722
},
{
"epoch": 0.25245392890144186,
"grad_norm": 0.20534907281398773,
"learning_rate": 5e-05,
"loss": 1.9933953285217285,
"step": 724
},
{
"epoch": 0.2531513154453685,
"grad_norm": 0.19531792402267456,
"learning_rate": 5e-05,
"loss": 2.0712335109710693,
"step": 726
},
{
"epoch": 0.2538487019892951,
"grad_norm": 0.2692209780216217,
"learning_rate": 5e-05,
"loss": 1.8906567096710205,
"step": 728
},
{
"epoch": 0.25454608853322175,
"grad_norm": 0.28854358196258545,
"learning_rate": 5e-05,
"loss": 2.0448617935180664,
"step": 730
},
{
"epoch": 0.2552434750771484,
"grad_norm": 0.21004259586334229,
"learning_rate": 5e-05,
"loss": 2.061755418777466,
"step": 732
},
{
"epoch": 0.25594086162107504,
"grad_norm": 0.1910502016544342,
"learning_rate": 5e-05,
"loss": 2.150097131729126,
"step": 734
},
{
"epoch": 0.25663824816500164,
"grad_norm": 0.24876752495765686,
"learning_rate": 5e-05,
"loss": 2.1157419681549072,
"step": 736
},
{
"epoch": 0.2573356347089283,
"grad_norm": 0.22561496496200562,
"learning_rate": 5e-05,
"loss": 2.087238311767578,
"step": 738
},
{
"epoch": 0.25803302125285493,
"grad_norm": 0.2667919993400574,
"learning_rate": 5e-05,
"loss": 2.117800712585449,
"step": 740
},
{
"epoch": 0.2587304077967816,
"grad_norm": 0.33913540840148926,
"learning_rate": 5e-05,
"loss": 2.0394740104675293,
"step": 742
},
{
"epoch": 0.2594277943407082,
"grad_norm": 0.1690003126859665,
"learning_rate": 5e-05,
"loss": 2.0725057125091553,
"step": 744
},
{
"epoch": 0.2601251808846348,
"grad_norm": 0.2350507229566574,
"learning_rate": 5e-05,
"loss": 2.180880546569824,
"step": 746
},
{
"epoch": 0.26082256742856147,
"grad_norm": 0.26825854182243347,
"learning_rate": 5e-05,
"loss": 2.1213607788085938,
"step": 748
},
{
"epoch": 0.2615199539724881,
"grad_norm": 0.1949106603860855,
"learning_rate": 5e-05,
"loss": 2.0407943725585938,
"step": 750
},
{
"epoch": 0.2622173405164147,
"grad_norm": 0.3467162549495697,
"learning_rate": 5e-05,
"loss": 1.9864469766616821,
"step": 752
},
{
"epoch": 0.26291472706034136,
"grad_norm": 0.20342646539211273,
"learning_rate": 5e-05,
"loss": 2.171985387802124,
"step": 754
},
{
"epoch": 0.263612113604268,
"grad_norm": 0.209679514169693,
"learning_rate": 5e-05,
"loss": 2.0921454429626465,
"step": 756
},
{
"epoch": 0.26430950014819465,
"grad_norm": 0.237525075674057,
"learning_rate": 5e-05,
"loss": 2.1939406394958496,
"step": 758
},
{
"epoch": 0.2650068866921213,
"grad_norm": 0.2145080417394638,
"learning_rate": 5e-05,
"loss": 2.132781505584717,
"step": 760
},
{
"epoch": 0.2657042732360479,
"grad_norm": 0.2128620594739914,
"learning_rate": 5e-05,
"loss": 2.053044080734253,
"step": 762
},
{
"epoch": 0.26640165977997454,
"grad_norm": 0.213362455368042,
"learning_rate": 5e-05,
"loss": 2.026813507080078,
"step": 764
},
{
"epoch": 0.2670990463239012,
"grad_norm": 0.28608018159866333,
"learning_rate": 5e-05,
"loss": 2.0595083236694336,
"step": 766
},
{
"epoch": 0.26779643286782784,
"grad_norm": 0.2690437138080597,
"learning_rate": 5e-05,
"loss": 1.9456102848052979,
"step": 768
},
{
"epoch": 0.26849381941175443,
"grad_norm": 0.16010865569114685,
"learning_rate": 5e-05,
"loss": 2.2088522911071777,
"step": 770
},
{
"epoch": 0.2691912059556811,
"grad_norm": 0.2786443531513214,
"learning_rate": 5e-05,
"loss": 2.180513381958008,
"step": 772
},
{
"epoch": 0.2698885924996077,
"grad_norm": 0.1664266735315323,
"learning_rate": 5e-05,
"loss": 2.1334125995635986,
"step": 774
},
{
"epoch": 0.2705859790435344,
"grad_norm": 0.21176394820213318,
"learning_rate": 5e-05,
"loss": 2.072683334350586,
"step": 776
},
{
"epoch": 0.27128336558746097,
"grad_norm": 0.2954282760620117,
"learning_rate": 5e-05,
"loss": 1.9445130825042725,
"step": 778
},
{
"epoch": 0.2719807521313876,
"grad_norm": 0.3181959092617035,
"learning_rate": 5e-05,
"loss": 2.138401508331299,
"step": 780
},
{
"epoch": 0.27267813867531426,
"grad_norm": 0.34124651551246643,
"learning_rate": 5e-05,
"loss": 2.053771495819092,
"step": 782
},
{
"epoch": 0.2733755252192409,
"grad_norm": 0.18755890429019928,
"learning_rate": 5e-05,
"loss": 2.2385504245758057,
"step": 784
},
{
"epoch": 0.2740729117631675,
"grad_norm": 0.23205752670764923,
"learning_rate": 5e-05,
"loss": 1.8543893098831177,
"step": 786
},
{
"epoch": 0.27477029830709415,
"grad_norm": 0.2637777626514435,
"learning_rate": 5e-05,
"loss": 2.216790199279785,
"step": 788
},
{
"epoch": 0.2754676848510208,
"grad_norm": 0.23640374839305878,
"learning_rate": 5e-05,
"loss": 1.9534659385681152,
"step": 790
},
{
"epoch": 0.27616507139494745,
"grad_norm": 0.3152559697628021,
"learning_rate": 5e-05,
"loss": 1.9674100875854492,
"step": 792
},
{
"epoch": 0.2768624579388741,
"grad_norm": 0.24841763079166412,
"learning_rate": 5e-05,
"loss": 1.9673519134521484,
"step": 794
},
{
"epoch": 0.2775598444828007,
"grad_norm": 0.2557215392589569,
"learning_rate": 5e-05,
"loss": 2.1061301231384277,
"step": 796
},
{
"epoch": 0.27825723102672734,
"grad_norm": 0.26791617274284363,
"learning_rate": 5e-05,
"loss": 1.980527639389038,
"step": 798
},
{
"epoch": 0.278954617570654,
"grad_norm": 0.18597114086151123,
"learning_rate": 5e-05,
"loss": 2.1571919918060303,
"step": 800
},
{
"epoch": 0.27965200411458063,
"grad_norm": 0.24104788899421692,
"learning_rate": 5e-05,
"loss": 2.052887439727783,
"step": 802
},
{
"epoch": 0.2803493906585072,
"grad_norm": 0.2629118263721466,
"learning_rate": 5e-05,
"loss": 2.1480886936187744,
"step": 804
},
{
"epoch": 0.2810467772024339,
"grad_norm": 0.1891702115535736,
"learning_rate": 5e-05,
"loss": 1.8945896625518799,
"step": 806
},
{
"epoch": 0.2817441637463605,
"grad_norm": 0.23194870352745056,
"learning_rate": 5e-05,
"loss": 2.0242464542388916,
"step": 808
},
{
"epoch": 0.28244155029028717,
"grad_norm": 0.22819095849990845,
"learning_rate": 5e-05,
"loss": 1.9482749700546265,
"step": 810
},
{
"epoch": 0.28313893683421376,
"grad_norm": 0.20101813971996307,
"learning_rate": 5e-05,
"loss": 2.1300370693206787,
"step": 812
},
{
"epoch": 0.2838363233781404,
"grad_norm": 0.371175080537796,
"learning_rate": 5e-05,
"loss": 2.1477580070495605,
"step": 814
},
{
"epoch": 0.28453370992206706,
"grad_norm": 0.30335918068885803,
"learning_rate": 5e-05,
"loss": 1.829369068145752,
"step": 816
},
{
"epoch": 0.2852310964659937,
"grad_norm": 0.2618269920349121,
"learning_rate": 5e-05,
"loss": 1.8751198053359985,
"step": 818
},
{
"epoch": 0.2859284830099203,
"grad_norm": 0.2387172430753708,
"learning_rate": 5e-05,
"loss": 1.9326030015945435,
"step": 820
},
{
"epoch": 0.28662586955384695,
"grad_norm": 0.24157649278640747,
"learning_rate": 5e-05,
"loss": 1.9281904697418213,
"step": 822
},
{
"epoch": 0.2873232560977736,
"grad_norm": 0.2316661924123764,
"learning_rate": 5e-05,
"loss": 2.1554112434387207,
"step": 824
},
{
"epoch": 0.28802064264170024,
"grad_norm": 0.3204042613506317,
"learning_rate": 5e-05,
"loss": 1.948148488998413,
"step": 826
},
{
"epoch": 0.2887180291856269,
"grad_norm": 0.20631511509418488,
"learning_rate": 5e-05,
"loss": 2.2995004653930664,
"step": 828
},
{
"epoch": 0.2894154157295535,
"grad_norm": 0.28268465399742126,
"learning_rate": 5e-05,
"loss": 2.0452728271484375,
"step": 830
},
{
"epoch": 0.29011280227348013,
"grad_norm": 0.2298143059015274,
"learning_rate": 5e-05,
"loss": 2.0092597007751465,
"step": 832
},
{
"epoch": 0.2908101888174068,
"grad_norm": 0.184764564037323,
"learning_rate": 5e-05,
"loss": 1.9803777933120728,
"step": 834
},
{
"epoch": 0.2915075753613334,
"grad_norm": 0.21297743916511536,
"learning_rate": 5e-05,
"loss": 2.1882874965667725,
"step": 836
},
{
"epoch": 0.29220496190526,
"grad_norm": 0.26148441433906555,
"learning_rate": 5e-05,
"loss": 1.9838454723358154,
"step": 838
},
{
"epoch": 0.29290234844918667,
"grad_norm": 0.25849202275276184,
"learning_rate": 5e-05,
"loss": 2.170018196105957,
"step": 840
},
{
"epoch": 0.2935997349931133,
"grad_norm": 0.19042696058750153,
"learning_rate": 5e-05,
"loss": 2.167074203491211,
"step": 842
},
{
"epoch": 0.29429712153703996,
"grad_norm": 0.21104653179645538,
"learning_rate": 5e-05,
"loss": 2.199953079223633,
"step": 844
},
{
"epoch": 0.29499450808096656,
"grad_norm": 0.2772143483161926,
"learning_rate": 5e-05,
"loss": 2.1958885192871094,
"step": 846
},
{
"epoch": 0.2956918946248932,
"grad_norm": 0.3362143337726593,
"learning_rate": 5e-05,
"loss": 1.926414966583252,
"step": 848
},
{
"epoch": 0.29638928116881985,
"grad_norm": 0.4233688712120056,
"learning_rate": 5e-05,
"loss": 1.7717297077178955,
"step": 850
},
{
"epoch": 0.2970866677127465,
"grad_norm": 0.33860331773757935,
"learning_rate": 5e-05,
"loss": 1.8843942880630493,
"step": 852
},
{
"epoch": 0.2977840542566731,
"grad_norm": 0.3185754418373108,
"learning_rate": 5e-05,
"loss": 2.1160013675689697,
"step": 854
},
{
"epoch": 0.29848144080059974,
"grad_norm": 0.20009522140026093,
"learning_rate": 5e-05,
"loss": 1.9877500534057617,
"step": 856
},
{
"epoch": 0.2991788273445264,
"grad_norm": 0.2403642237186432,
"learning_rate": 5e-05,
"loss": 2.1462478637695312,
"step": 858
},
{
"epoch": 0.29987621388845304,
"grad_norm": 0.21490423381328583,
"learning_rate": 5e-05,
"loss": 2.0985560417175293,
"step": 860
},
{
"epoch": 0.3005736004323797,
"grad_norm": 0.2281097024679184,
"learning_rate": 5e-05,
"loss": 1.953871488571167,
"step": 862
},
{
"epoch": 0.3012709869763063,
"grad_norm": 0.22519268095493317,
"learning_rate": 5e-05,
"loss": 1.9938616752624512,
"step": 864
},
{
"epoch": 0.3019683735202329,
"grad_norm": 0.2882140576839447,
"learning_rate": 5e-05,
"loss": 2.0265111923217773,
"step": 866
},
{
"epoch": 0.3026657600641596,
"grad_norm": 0.26522985100746155,
"learning_rate": 5e-05,
"loss": 1.971495270729065,
"step": 868
},
{
"epoch": 0.3033631466080862,
"grad_norm": 0.2441837340593338,
"learning_rate": 5e-05,
"loss": 1.90657639503479,
"step": 870
},
{
"epoch": 0.3040605331520128,
"grad_norm": 0.33309900760650635,
"learning_rate": 5e-05,
"loss": 2.001520872116089,
"step": 872
},
{
"epoch": 0.30475791969593946,
"grad_norm": 0.2041439563035965,
"learning_rate": 5e-05,
"loss": 1.968753695487976,
"step": 874
},
{
"epoch": 0.3054553062398661,
"grad_norm": 0.17259293794631958,
"learning_rate": 5e-05,
"loss": 1.9523271322250366,
"step": 876
},
{
"epoch": 0.30615269278379276,
"grad_norm": 0.1741812825202942,
"learning_rate": 5e-05,
"loss": 2.1121654510498047,
"step": 878
},
{
"epoch": 0.30685007932771935,
"grad_norm": 0.23253613710403442,
"learning_rate": 5e-05,
"loss": 2.213024616241455,
"step": 880
},
{
"epoch": 0.307547465871646,
"grad_norm": 0.2259003072977066,
"learning_rate": 5e-05,
"loss": 2.094529151916504,
"step": 882
},
{
"epoch": 0.30824485241557265,
"grad_norm": 0.22867822647094727,
"learning_rate": 5e-05,
"loss": 2.0620737075805664,
"step": 884
},
{
"epoch": 0.3089422389594993,
"grad_norm": 0.17488710582256317,
"learning_rate": 5e-05,
"loss": 1.985865592956543,
"step": 886
},
{
"epoch": 0.3096396255034259,
"grad_norm": 0.28305572271347046,
"learning_rate": 5e-05,
"loss": 2.090651273727417,
"step": 888
},
{
"epoch": 0.31033701204735253,
"grad_norm": 0.258344829082489,
"learning_rate": 5e-05,
"loss": 1.9839575290679932,
"step": 890
},
{
"epoch": 0.3110343985912792,
"grad_norm": 0.2704688310623169,
"learning_rate": 5e-05,
"loss": 1.9838356971740723,
"step": 892
},
{
"epoch": 0.31173178513520583,
"grad_norm": 0.21489857137203217,
"learning_rate": 5e-05,
"loss": 2.0196244716644287,
"step": 894
},
{
"epoch": 0.3124291716791325,
"grad_norm": 0.42650023102760315,
"learning_rate": 5e-05,
"loss": 2.060972213745117,
"step": 896
},
{
"epoch": 0.31312655822305907,
"grad_norm": 0.17513501644134521,
"learning_rate": 5e-05,
"loss": 2.140475273132324,
"step": 898
},
{
"epoch": 0.3138239447669857,
"grad_norm": 0.2620815336704254,
"learning_rate": 5e-05,
"loss": 2.194441318511963,
"step": 900
},
{
"epoch": 0.31452133131091237,
"grad_norm": 0.2540329396724701,
"learning_rate": 5e-05,
"loss": 1.9437588453292847,
"step": 902
},
{
"epoch": 0.315218717854839,
"grad_norm": 0.19750094413757324,
"learning_rate": 5e-05,
"loss": 2.0057075023651123,
"step": 904
},
{
"epoch": 0.3159161043987656,
"grad_norm": 0.23293620347976685,
"learning_rate": 5e-05,
"loss": 1.8982203006744385,
"step": 906
},
{
"epoch": 0.31661349094269226,
"grad_norm": 0.2062148004770279,
"learning_rate": 5e-05,
"loss": 2.0358047485351562,
"step": 908
},
{
"epoch": 0.3173108774866189,
"grad_norm": 0.26289623975753784,
"learning_rate": 5e-05,
"loss": 1.795961618423462,
"step": 910
},
{
"epoch": 0.31800826403054555,
"grad_norm": 0.2925187349319458,
"learning_rate": 5e-05,
"loss": 1.9860413074493408,
"step": 912
},
{
"epoch": 0.31870565057447214,
"grad_norm": 0.2811063826084137,
"learning_rate": 5e-05,
"loss": 1.8662142753601074,
"step": 914
},
{
"epoch": 0.3194030371183988,
"grad_norm": 0.22942379117012024,
"learning_rate": 5e-05,
"loss": 2.0050673484802246,
"step": 916
},
{
"epoch": 0.32010042366232544,
"grad_norm": 0.16354793310165405,
"learning_rate": 5e-05,
"loss": 2.1945548057556152,
"step": 918
},
{
"epoch": 0.3207978102062521,
"grad_norm": 0.23471316695213318,
"learning_rate": 5e-05,
"loss": 2.2042596340179443,
"step": 920
},
{
"epoch": 0.3214951967501787,
"grad_norm": 0.2259097695350647,
"learning_rate": 5e-05,
"loss": 2.16628360748291,
"step": 922
},
{
"epoch": 0.32219258329410533,
"grad_norm": 0.2739982008934021,
"learning_rate": 5e-05,
"loss": 2.055208206176758,
"step": 924
},
{
"epoch": 0.322889969838032,
"grad_norm": 0.3012983798980713,
"learning_rate": 5e-05,
"loss": 2.2179489135742188,
"step": 926
},
{
"epoch": 0.3235873563819586,
"grad_norm": 0.23193614184856415,
"learning_rate": 5e-05,
"loss": 2.104721784591675,
"step": 928
},
{
"epoch": 0.32428474292588527,
"grad_norm": 0.19617155194282532,
"learning_rate": 5e-05,
"loss": 1.9976686239242554,
"step": 930
},
{
"epoch": 0.32498212946981186,
"grad_norm": 0.22112447023391724,
"learning_rate": 5e-05,
"loss": 2.2046685218811035,
"step": 932
},
{
"epoch": 0.3256795160137385,
"grad_norm": 0.19225233793258667,
"learning_rate": 5e-05,
"loss": 1.9811292886734009,
"step": 934
},
{
"epoch": 0.32637690255766516,
"grad_norm": 0.16594399511814117,
"learning_rate": 5e-05,
"loss": 2.0854458808898926,
"step": 936
},
{
"epoch": 0.3270742891015918,
"grad_norm": 0.22566556930541992,
"learning_rate": 5e-05,
"loss": 2.0371317863464355,
"step": 938
},
{
"epoch": 0.3277716756455184,
"grad_norm": 0.2621348798274994,
"learning_rate": 5e-05,
"loss": 2.2924721240997314,
"step": 940
},
{
"epoch": 0.32846906218944505,
"grad_norm": 0.3321608006954193,
"learning_rate": 5e-05,
"loss": 1.9144505262374878,
"step": 942
},
{
"epoch": 0.3291664487333717,
"grad_norm": 0.21470032632350922,
"learning_rate": 5e-05,
"loss": 2.0083987712860107,
"step": 944
},
{
"epoch": 0.32986383527729835,
"grad_norm": 0.17992348968982697,
"learning_rate": 5e-05,
"loss": 2.096499443054199,
"step": 946
},
{
"epoch": 0.33056122182122494,
"grad_norm": 0.191554456949234,
"learning_rate": 5e-05,
"loss": 1.956010103225708,
"step": 948
},
{
"epoch": 0.3312586083651516,
"grad_norm": 0.2133859097957611,
"learning_rate": 5e-05,
"loss": 1.908299446105957,
"step": 950
},
{
"epoch": 0.33195599490907823,
"grad_norm": 0.22485746443271637,
"learning_rate": 5e-05,
"loss": 1.7729823589324951,
"step": 952
},
{
"epoch": 0.3326533814530049,
"grad_norm": 0.186967670917511,
"learning_rate": 5e-05,
"loss": 1.9951062202453613,
"step": 954
},
{
"epoch": 0.3333507679969315,
"grad_norm": 0.26257404685020447,
"learning_rate": 5e-05,
"loss": 2.0230295658111572,
"step": 956
},
{
"epoch": 0.3340481545408581,
"grad_norm": 0.2617090046405792,
"learning_rate": 5e-05,
"loss": 1.9331642389297485,
"step": 958
},
{
"epoch": 0.33474554108478477,
"grad_norm": 0.20375195145606995,
"learning_rate": 5e-05,
"loss": 1.9368494749069214,
"step": 960
},
{
"epoch": 0.3354429276287114,
"grad_norm": 0.1918850988149643,
"learning_rate": 5e-05,
"loss": 2.1231679916381836,
"step": 962
},
{
"epoch": 0.33614031417263807,
"grad_norm": 0.26361358165740967,
"learning_rate": 5e-05,
"loss": 2.1133885383605957,
"step": 964
},
{
"epoch": 0.33683770071656466,
"grad_norm": 0.19686871767044067,
"learning_rate": 5e-05,
"loss": 2.07720685005188,
"step": 966
},
{
"epoch": 0.3375350872604913,
"grad_norm": 0.32117438316345215,
"learning_rate": 5e-05,
"loss": 2.0643956661224365,
"step": 968
},
{
"epoch": 0.33823247380441795,
"grad_norm": 0.2792845368385315,
"learning_rate": 5e-05,
"loss": 2.0637881755828857,
"step": 970
},
{
"epoch": 0.3389298603483446,
"grad_norm": 0.21377836167812347,
"learning_rate": 5e-05,
"loss": 2.150491714477539,
"step": 972
},
{
"epoch": 0.3396272468922712,
"grad_norm": 0.22636762261390686,
"learning_rate": 5e-05,
"loss": 1.9907886981964111,
"step": 974
},
{
"epoch": 0.34032463343619784,
"grad_norm": 0.2062670737504959,
"learning_rate": 5e-05,
"loss": 1.9640690088272095,
"step": 976
},
{
"epoch": 0.3410220199801245,
"grad_norm": 0.2347593754529953,
"learning_rate": 5e-05,
"loss": 2.0291075706481934,
"step": 978
},
{
"epoch": 0.34171940652405114,
"grad_norm": 0.22571015357971191,
"learning_rate": 5e-05,
"loss": 2.0671143531799316,
"step": 980
},
{
"epoch": 0.34241679306797773,
"grad_norm": 0.2145722508430481,
"learning_rate": 5e-05,
"loss": 2.1719367504119873,
"step": 982
},
{
"epoch": 0.3431141796119044,
"grad_norm": 0.25082701444625854,
"learning_rate": 5e-05,
"loss": 2.0923759937286377,
"step": 984
},
{
"epoch": 0.343811566155831,
"grad_norm": 0.22755537927150726,
"learning_rate": 5e-05,
"loss": 1.9666999578475952,
"step": 986
},
{
"epoch": 0.3445089526997577,
"grad_norm": 0.2901584804058075,
"learning_rate": 5e-05,
"loss": 1.8081202507019043,
"step": 988
},
{
"epoch": 0.34520633924368427,
"grad_norm": 0.23690833151340485,
"learning_rate": 5e-05,
"loss": 2.040344476699829,
"step": 990
},
{
"epoch": 0.3459037257876109,
"grad_norm": 0.21881553530693054,
"learning_rate": 5e-05,
"loss": 1.7107083797454834,
"step": 992
},
{
"epoch": 0.34660111233153756,
"grad_norm": 0.26182490587234497,
"learning_rate": 5e-05,
"loss": 2.064363479614258,
"step": 994
},
{
"epoch": 0.3472984988754642,
"grad_norm": 0.2080112099647522,
"learning_rate": 5e-05,
"loss": 2.166213274002075,
"step": 996
},
{
"epoch": 0.34799588541939086,
"grad_norm": 0.2145555168390274,
"learning_rate": 5e-05,
"loss": 1.9562971591949463,
"step": 998
},
{
"epoch": 0.34869327196331745,
"grad_norm": 0.21929831802845,
"learning_rate": 5e-05,
"loss": 2.0631940364837646,
"step": 1000
},
{
"epoch": 0.3493906585072441,
"grad_norm": 0.2061082273721695,
"learning_rate": 5e-05,
"loss": 1.9821693897247314,
"step": 1002
},
{
"epoch": 0.35008804505117075,
"grad_norm": 0.21569553017616272,
"learning_rate": 5e-05,
"loss": 2.1045546531677246,
"step": 1004
},
{
"epoch": 0.3507854315950974,
"grad_norm": 0.20195575058460236,
"learning_rate": 5e-05,
"loss": 2.023355484008789,
"step": 1006
},
{
"epoch": 0.351482818139024,
"grad_norm": 0.21589083969593048,
"learning_rate": 5e-05,
"loss": 1.9698195457458496,
"step": 1008
},
{
"epoch": 0.35218020468295064,
"grad_norm": 0.35657018423080444,
"learning_rate": 5e-05,
"loss": 2.0477797985076904,
"step": 1010
},
{
"epoch": 0.3528775912268773,
"grad_norm": 0.2049272358417511,
"learning_rate": 5e-05,
"loss": 2.010043144226074,
"step": 1012
},
{
"epoch": 0.35357497777080393,
"grad_norm": 0.1918814480304718,
"learning_rate": 5e-05,
"loss": 1.9827858209609985,
"step": 1014
},
{
"epoch": 0.3542723643147305,
"grad_norm": 0.22987155616283417,
"learning_rate": 5e-05,
"loss": 1.987635612487793,
"step": 1016
},
{
"epoch": 0.3549697508586572,
"grad_norm": 0.24195624887943268,
"learning_rate": 5e-05,
"loss": 1.9928193092346191,
"step": 1018
},
{
"epoch": 0.3556671374025838,
"grad_norm": 0.27684086561203003,
"learning_rate": 5e-05,
"loss": 1.9530699253082275,
"step": 1020
},
{
"epoch": 0.35636452394651047,
"grad_norm": 0.21984905004501343,
"learning_rate": 5e-05,
"loss": 2.0691213607788086,
"step": 1022
},
{
"epoch": 0.35706191049043706,
"grad_norm": 0.249387726187706,
"learning_rate": 5e-05,
"loss": 2.183511734008789,
"step": 1024
},
{
"epoch": 0.3577592970343637,
"grad_norm": 0.23080800473690033,
"learning_rate": 5e-05,
"loss": 1.930666446685791,
"step": 1026
},
{
"epoch": 0.35845668357829036,
"grad_norm": 0.23000819981098175,
"learning_rate": 5e-05,
"loss": 2.092984199523926,
"step": 1028
},
{
"epoch": 0.359154070122217,
"grad_norm": 0.1848149597644806,
"learning_rate": 5e-05,
"loss": 2.0390803813934326,
"step": 1030
},
{
"epoch": 0.35985145666614365,
"grad_norm": 0.27325889468193054,
"learning_rate": 5e-05,
"loss": 1.9925591945648193,
"step": 1032
},
{
"epoch": 0.36054884321007025,
"grad_norm": 0.2851124405860901,
"learning_rate": 5e-05,
"loss": 2.2212448120117188,
"step": 1034
},
{
"epoch": 0.3612462297539969,
"grad_norm": 0.357220321893692,
"learning_rate": 5e-05,
"loss": 1.5241745710372925,
"step": 1036
},
{
"epoch": 0.36194361629792354,
"grad_norm": 0.28451353311538696,
"learning_rate": 5e-05,
"loss": 2.13199782371521,
"step": 1038
},
{
"epoch": 0.3626410028418502,
"grad_norm": 0.17099227011203766,
"learning_rate": 5e-05,
"loss": 1.9190986156463623,
"step": 1040
},
{
"epoch": 0.3633383893857768,
"grad_norm": 0.27135729789733887,
"learning_rate": 5e-05,
"loss": 2.1251871585845947,
"step": 1042
},
{
"epoch": 0.36403577592970343,
"grad_norm": 0.28680723905563354,
"learning_rate": 5e-05,
"loss": 1.6807236671447754,
"step": 1044
},
{
"epoch": 0.3647331624736301,
"grad_norm": 0.2310262769460678,
"learning_rate": 5e-05,
"loss": 2.03737735748291,
"step": 1046
},
{
"epoch": 0.3654305490175567,
"grad_norm": 0.19282707571983337,
"learning_rate": 5e-05,
"loss": 2.1107892990112305,
"step": 1048
},
{
"epoch": 0.3661279355614833,
"grad_norm": 0.22267811000347137,
"learning_rate": 5e-05,
"loss": 1.990861415863037,
"step": 1050
},
{
"epoch": 0.36682532210540997,
"grad_norm": 0.23680366575717926,
"learning_rate": 5e-05,
"loss": 1.9701367616653442,
"step": 1052
},
{
"epoch": 0.3675227086493366,
"grad_norm": 0.2254926711320877,
"learning_rate": 5e-05,
"loss": 1.8134162425994873,
"step": 1054
},
{
"epoch": 0.36822009519326326,
"grad_norm": 0.22644150257110596,
"learning_rate": 5e-05,
"loss": 1.911505937576294,
"step": 1056
},
{
"epoch": 0.36891748173718986,
"grad_norm": 0.1897391378879547,
"learning_rate": 5e-05,
"loss": 2.0960769653320312,
"step": 1058
},
{
"epoch": 0.3696148682811165,
"grad_norm": 0.26435860991477966,
"learning_rate": 5e-05,
"loss": 2.0274524688720703,
"step": 1060
},
{
"epoch": 0.37031225482504315,
"grad_norm": 0.33470162749290466,
"learning_rate": 5e-05,
"loss": 1.8201732635498047,
"step": 1062
},
{
"epoch": 0.3710096413689698,
"grad_norm": 0.22211971879005432,
"learning_rate": 5e-05,
"loss": 1.9741380214691162,
"step": 1064
},
{
"epoch": 0.37170702791289645,
"grad_norm": 0.38349875807762146,
"learning_rate": 5e-05,
"loss": 2.1487531661987305,
"step": 1066
},
{
"epoch": 0.37240441445682304,
"grad_norm": 0.2047680765390396,
"learning_rate": 5e-05,
"loss": 2.2522268295288086,
"step": 1068
},
{
"epoch": 0.3731018010007497,
"grad_norm": 0.2490031123161316,
"learning_rate": 5e-05,
"loss": 2.051534652709961,
"step": 1070
},
{
"epoch": 0.37379918754467634,
"grad_norm": 0.22337035834789276,
"learning_rate": 5e-05,
"loss": 2.099733829498291,
"step": 1072
},
{
"epoch": 0.374496574088603,
"grad_norm": 0.22020231187343597,
"learning_rate": 5e-05,
"loss": 2.078166961669922,
"step": 1074
},
{
"epoch": 0.3751939606325296,
"grad_norm": 0.26717209815979004,
"learning_rate": 5e-05,
"loss": 1.8090195655822754,
"step": 1076
},
{
"epoch": 0.3758913471764562,
"grad_norm": 0.1886700689792633,
"learning_rate": 5e-05,
"loss": 2.0888776779174805,
"step": 1078
},
{
"epoch": 0.3765887337203829,
"grad_norm": 0.2425251603126526,
"learning_rate": 5e-05,
"loss": 1.9623768329620361,
"step": 1080
},
{
"epoch": 0.3772861202643095,
"grad_norm": 0.22117263078689575,
"learning_rate": 5e-05,
"loss": 2.085573196411133,
"step": 1082
},
{
"epoch": 0.3779835068082361,
"grad_norm": 0.2838881015777588,
"learning_rate": 5e-05,
"loss": 2.0469043254852295,
"step": 1084
},
{
"epoch": 0.37868089335216276,
"grad_norm": 0.1819688081741333,
"learning_rate": 5e-05,
"loss": 1.5593383312225342,
"step": 1086
},
{
"epoch": 0.3793782798960894,
"grad_norm": 0.2294865846633911,
"learning_rate": 5e-05,
"loss": 1.9427132606506348,
"step": 1088
},
{
"epoch": 0.38007566644001606,
"grad_norm": 0.23989562690258026,
"learning_rate": 5e-05,
"loss": 1.9454470872879028,
"step": 1090
},
{
"epoch": 0.38077305298394265,
"grad_norm": 0.31190624833106995,
"learning_rate": 5e-05,
"loss": 2.0495333671569824,
"step": 1092
},
{
"epoch": 0.3814704395278693,
"grad_norm": 0.1790897399187088,
"learning_rate": 5e-05,
"loss": 2.115795850753784,
"step": 1094
},
{
"epoch": 0.38216782607179595,
"grad_norm": 0.23868292570114136,
"learning_rate": 5e-05,
"loss": 2.1075124740600586,
"step": 1096
},
{
"epoch": 0.3828652126157226,
"grad_norm": 0.2061663120985031,
"learning_rate": 5e-05,
"loss": 1.8393436670303345,
"step": 1098
},
{
"epoch": 0.3835625991596492,
"grad_norm": 0.18991388380527496,
"learning_rate": 5e-05,
"loss": 2.1416378021240234,
"step": 1100
},
{
"epoch": 0.38425998570357583,
"grad_norm": 0.21696053445339203,
"learning_rate": 5e-05,
"loss": 2.0470800399780273,
"step": 1102
},
{
"epoch": 0.3849573722475025,
"grad_norm": 0.3593919277191162,
"learning_rate": 5e-05,
"loss": 1.9083117246627808,
"step": 1104
},
{
"epoch": 0.38565475879142913,
"grad_norm": 0.17209303379058838,
"learning_rate": 5e-05,
"loss": 2.0877814292907715,
"step": 1106
},
{
"epoch": 0.3863521453353558,
"grad_norm": 0.22013252973556519,
"learning_rate": 5e-05,
"loss": 2.1486687660217285,
"step": 1108
},
{
"epoch": 0.38704953187928237,
"grad_norm": 0.1675334870815277,
"learning_rate": 5e-05,
"loss": 2.065140724182129,
"step": 1110
},
{
"epoch": 0.387746918423209,
"grad_norm": 0.2227640151977539,
"learning_rate": 5e-05,
"loss": 1.9250330924987793,
"step": 1112
},
{
"epoch": 0.38844430496713567,
"grad_norm": 0.21977634727954865,
"learning_rate": 5e-05,
"loss": 1.8304221630096436,
"step": 1114
},
{
"epoch": 0.3891416915110623,
"grad_norm": 0.21973222494125366,
"learning_rate": 5e-05,
"loss": 1.9635357856750488,
"step": 1116
},
{
"epoch": 0.3898390780549889,
"grad_norm": 0.2314382642507553,
"learning_rate": 5e-05,
"loss": 1.9372278451919556,
"step": 1118
},
{
"epoch": 0.39053646459891556,
"grad_norm": 0.21442043781280518,
"learning_rate": 5e-05,
"loss": 1.7821249961853027,
"step": 1120
},
{
"epoch": 0.3912338511428422,
"grad_norm": 0.22008800506591797,
"learning_rate": 5e-05,
"loss": 2.0237293243408203,
"step": 1122
},
{
"epoch": 0.39193123768676885,
"grad_norm": 0.23807160556316376,
"learning_rate": 5e-05,
"loss": 2.054992198944092,
"step": 1124
},
{
"epoch": 0.39262862423069544,
"grad_norm": 0.437844842672348,
"learning_rate": 5e-05,
"loss": 1.7409727573394775,
"step": 1126
},
{
"epoch": 0.3933260107746221,
"grad_norm": 0.21472229063510895,
"learning_rate": 5e-05,
"loss": 2.0891048908233643,
"step": 1128
},
{
"epoch": 0.39402339731854874,
"grad_norm": 0.38082173466682434,
"learning_rate": 5e-05,
"loss": 1.919767141342163,
"step": 1130
},
{
"epoch": 0.3947207838624754,
"grad_norm": 0.3429301083087921,
"learning_rate": 5e-05,
"loss": 1.9058270454406738,
"step": 1132
},
{
"epoch": 0.395418170406402,
"grad_norm": 0.22207891941070557,
"learning_rate": 5e-05,
"loss": 2.0372676849365234,
"step": 1134
},
{
"epoch": 0.39611555695032863,
"grad_norm": 0.2745357155799866,
"learning_rate": 5e-05,
"loss": 1.9663827419281006,
"step": 1136
},
{
"epoch": 0.3968129434942553,
"grad_norm": 0.3773990869522095,
"learning_rate": 5e-05,
"loss": 1.9896197319030762,
"step": 1138
},
{
"epoch": 0.3975103300381819,
"grad_norm": 0.2900267243385315,
"learning_rate": 5e-05,
"loss": 1.7686326503753662,
"step": 1140
},
{
"epoch": 0.3982077165821086,
"grad_norm": 0.22188392281532288,
"learning_rate": 5e-05,
"loss": 2.0737409591674805,
"step": 1142
},
{
"epoch": 0.39890510312603517,
"grad_norm": 0.2207290530204773,
"learning_rate": 5e-05,
"loss": 2.016484260559082,
"step": 1144
},
{
"epoch": 0.3996024896699618,
"grad_norm": 0.1925177276134491,
"learning_rate": 5e-05,
"loss": 1.7681307792663574,
"step": 1146
},
{
"epoch": 0.40029987621388846,
"grad_norm": 0.19402268528938293,
"learning_rate": 5e-05,
"loss": 2.1917271614074707,
"step": 1148
},
{
"epoch": 0.4009972627578151,
"grad_norm": 0.2637289762496948,
"learning_rate": 5e-05,
"loss": 2.018184185028076,
"step": 1150
},
{
"epoch": 0.4016946493017417,
"grad_norm": 0.23698849976062775,
"learning_rate": 5e-05,
"loss": 1.897341012954712,
"step": 1152
},
{
"epoch": 0.40239203584566835,
"grad_norm": 0.2660623788833618,
"learning_rate": 5e-05,
"loss": 2.0587849617004395,
"step": 1154
},
{
"epoch": 0.403089422389595,
"grad_norm": 0.19925248622894287,
"learning_rate": 5e-05,
"loss": 2.0214662551879883,
"step": 1156
},
{
"epoch": 0.40378680893352165,
"grad_norm": 0.24144220352172852,
"learning_rate": 5e-05,
"loss": 2.018463611602783,
"step": 1158
},
{
"epoch": 0.40448419547744824,
"grad_norm": 0.1997368037700653,
"learning_rate": 5e-05,
"loss": 2.16738224029541,
"step": 1160
},
{
"epoch": 0.4051815820213749,
"grad_norm": 0.2711610794067383,
"learning_rate": 5e-05,
"loss": 2.0850989818573,
"step": 1162
},
{
"epoch": 0.40587896856530153,
"grad_norm": 0.23238755762577057,
"learning_rate": 5e-05,
"loss": 1.8553071022033691,
"step": 1164
},
{
"epoch": 0.4065763551092282,
"grad_norm": 0.19778624176979065,
"learning_rate": 5e-05,
"loss": 2.267437696456909,
"step": 1166
},
{
"epoch": 0.4072737416531548,
"grad_norm": 0.18753691017627716,
"learning_rate": 5e-05,
"loss": 1.9425708055496216,
"step": 1168
},
{
"epoch": 0.4079711281970814,
"grad_norm": 0.1919858753681183,
"learning_rate": 5e-05,
"loss": 2.0795068740844727,
"step": 1170
},
{
"epoch": 0.40866851474100807,
"grad_norm": 0.2890529930591583,
"learning_rate": 5e-05,
"loss": 2.067218780517578,
"step": 1172
},
{
"epoch": 0.4093659012849347,
"grad_norm": 0.28386932611465454,
"learning_rate": 5e-05,
"loss": 2.009732723236084,
"step": 1174
},
{
"epoch": 0.41006328782886137,
"grad_norm": 0.22415214776992798,
"learning_rate": 5e-05,
"loss": 2.1026344299316406,
"step": 1176
},
{
"epoch": 0.41076067437278796,
"grad_norm": 0.2018672227859497,
"learning_rate": 5e-05,
"loss": 2.120260715484619,
"step": 1178
},
{
"epoch": 0.4114580609167146,
"grad_norm": 0.2746317386627197,
"learning_rate": 5e-05,
"loss": 1.9730490446090698,
"step": 1180
},
{
"epoch": 0.41215544746064126,
"grad_norm": 0.24063794314861298,
"learning_rate": 5e-05,
"loss": 2.0942440032958984,
"step": 1182
},
{
"epoch": 0.4128528340045679,
"grad_norm": 0.24473215639591217,
"learning_rate": 5e-05,
"loss": 2.0083141326904297,
"step": 1184
},
{
"epoch": 0.4135502205484945,
"grad_norm": 0.27954408526420593,
"learning_rate": 5e-05,
"loss": 1.9019955396652222,
"step": 1186
},
{
"epoch": 0.41424760709242114,
"grad_norm": 0.21032820641994476,
"learning_rate": 5e-05,
"loss": 2.0116095542907715,
"step": 1188
},
{
"epoch": 0.4149449936363478,
"grad_norm": 0.29512640833854675,
"learning_rate": 5e-05,
"loss": 2.1200222969055176,
"step": 1190
},
{
"epoch": 0.41564238018027444,
"grad_norm": 0.184640035033226,
"learning_rate": 5e-05,
"loss": 2.117198944091797,
"step": 1192
},
{
"epoch": 0.41633976672420103,
"grad_norm": 0.2741994559764862,
"learning_rate": 5e-05,
"loss": 1.9714668989181519,
"step": 1194
},
{
"epoch": 0.4170371532681277,
"grad_norm": 0.3104700446128845,
"learning_rate": 5e-05,
"loss": 1.955941081047058,
"step": 1196
},
{
"epoch": 0.41773453981205433,
"grad_norm": 0.25845324993133545,
"learning_rate": 5e-05,
"loss": 2.067126512527466,
"step": 1198
},
{
"epoch": 0.418431926355981,
"grad_norm": 0.16517043113708496,
"learning_rate": 5e-05,
"loss": 2.07180118560791,
"step": 1200
},
{
"epoch": 0.41912931289990757,
"grad_norm": 0.15481635928153992,
"learning_rate": 5e-05,
"loss": 2.18418550491333,
"step": 1202
},
{
"epoch": 0.4198266994438342,
"grad_norm": 0.2163589596748352,
"learning_rate": 5e-05,
"loss": 2.1593503952026367,
"step": 1204
},
{
"epoch": 0.42052408598776087,
"grad_norm": 0.2110627144575119,
"learning_rate": 5e-05,
"loss": 1.9078938961029053,
"step": 1206
},
{
"epoch": 0.4212214725316875,
"grad_norm": 0.2240639328956604,
"learning_rate": 5e-05,
"loss": 1.9624361991882324,
"step": 1208
},
{
"epoch": 0.42191885907561416,
"grad_norm": 0.18361057341098785,
"learning_rate": 5e-05,
"loss": 2.1403841972351074,
"step": 1210
},
{
"epoch": 0.42261624561954075,
"grad_norm": 0.29115167260169983,
"learning_rate": 5e-05,
"loss": 2.0681142807006836,
"step": 1212
},
{
"epoch": 0.4233136321634674,
"grad_norm": 0.3003404140472412,
"learning_rate": 5e-05,
"loss": 2.1037189960479736,
"step": 1214
},
{
"epoch": 0.42401101870739405,
"grad_norm": 0.2676253020763397,
"learning_rate": 5e-05,
"loss": 2.052903175354004,
"step": 1216
},
{
"epoch": 0.4247084052513207,
"grad_norm": 0.19431860744953156,
"learning_rate": 5e-05,
"loss": 1.8593212366104126,
"step": 1218
},
{
"epoch": 0.4254057917952473,
"grad_norm": 0.2959866523742676,
"learning_rate": 5e-05,
"loss": 2.089138984680176,
"step": 1220
},
{
"epoch": 0.42610317833917394,
"grad_norm": 0.21319754421710968,
"learning_rate": 5e-05,
"loss": 2.118983268737793,
"step": 1222
},
{
"epoch": 0.4268005648831006,
"grad_norm": 0.2284935712814331,
"learning_rate": 5e-05,
"loss": 2.0338568687438965,
"step": 1224
},
{
"epoch": 0.42749795142702723,
"grad_norm": 0.20905207097530365,
"learning_rate": 5e-05,
"loss": 2.126631498336792,
"step": 1226
},
{
"epoch": 0.4281953379709538,
"grad_norm": 0.21172241866588593,
"learning_rate": 5e-05,
"loss": 2.0528483390808105,
"step": 1228
},
{
"epoch": 0.4288927245148805,
"grad_norm": 0.2142503708600998,
"learning_rate": 5e-05,
"loss": 2.0093178749084473,
"step": 1230
},
{
"epoch": 0.4295901110588071,
"grad_norm": 0.2018240988254547,
"learning_rate": 5e-05,
"loss": 2.0821194648742676,
"step": 1232
},
{
"epoch": 0.43028749760273377,
"grad_norm": 0.26273906230926514,
"learning_rate": 5e-05,
"loss": 2.0322070121765137,
"step": 1234
},
{
"epoch": 0.43098488414666036,
"grad_norm": 0.2783421277999878,
"learning_rate": 5e-05,
"loss": 2.0080924034118652,
"step": 1236
},
{
"epoch": 0.431682270690587,
"grad_norm": 0.1750483363866806,
"learning_rate": 5e-05,
"loss": 2.006192684173584,
"step": 1238
},
{
"epoch": 0.43237965723451366,
"grad_norm": 0.2582997679710388,
"learning_rate": 5e-05,
"loss": 1.858557105064392,
"step": 1240
},
{
"epoch": 0.4330770437784403,
"grad_norm": 0.17965590953826904,
"learning_rate": 5e-05,
"loss": 2.3109617233276367,
"step": 1242
},
{
"epoch": 0.43377443032236696,
"grad_norm": 0.21540866792201996,
"learning_rate": 5e-05,
"loss": 2.129401683807373,
"step": 1244
},
{
"epoch": 0.43447181686629355,
"grad_norm": 0.21578910946846008,
"learning_rate": 5e-05,
"loss": 2.101186752319336,
"step": 1246
},
{
"epoch": 0.4351692034102202,
"grad_norm": 0.3152891993522644,
"learning_rate": 5e-05,
"loss": 1.7350882291793823,
"step": 1248
},
{
"epoch": 0.43586658995414684,
"grad_norm": 0.1738380789756775,
"learning_rate": 5e-05,
"loss": 1.8759608268737793,
"step": 1250
},
{
"epoch": 0.4365639764980735,
"grad_norm": 0.1779043972492218,
"learning_rate": 5e-05,
"loss": 1.8695545196533203,
"step": 1252
},
{
"epoch": 0.4372613630420001,
"grad_norm": 0.22195081412792206,
"learning_rate": 5e-05,
"loss": 1.7950637340545654,
"step": 1254
},
{
"epoch": 0.43795874958592673,
"grad_norm": 0.19365878403186798,
"learning_rate": 5e-05,
"loss": 2.0437979698181152,
"step": 1256
},
{
"epoch": 0.4386561361298534,
"grad_norm": 0.15562410652637482,
"learning_rate": 5e-05,
"loss": 1.8667664527893066,
"step": 1258
},
{
"epoch": 0.43935352267378003,
"grad_norm": 0.2590292692184448,
"learning_rate": 5e-05,
"loss": 1.9112123250961304,
"step": 1260
},
{
"epoch": 0.4400509092177066,
"grad_norm": 0.25530561804771423,
"learning_rate": 5e-05,
"loss": 1.7089579105377197,
"step": 1262
},
{
"epoch": 0.44074829576163327,
"grad_norm": 0.18177051842212677,
"learning_rate": 5e-05,
"loss": 2.036484956741333,
"step": 1264
},
{
"epoch": 0.4414456823055599,
"grad_norm": 0.23425745964050293,
"learning_rate": 5e-05,
"loss": 2.0025293827056885,
"step": 1266
},
{
"epoch": 0.44214306884948656,
"grad_norm": 0.18502822518348694,
"learning_rate": 5e-05,
"loss": 1.8278251886367798,
"step": 1268
},
{
"epoch": 0.44284045539341316,
"grad_norm": 0.3032926917076111,
"learning_rate": 5e-05,
"loss": 1.9995391368865967,
"step": 1270
},
{
"epoch": 0.4435378419373398,
"grad_norm": 0.19650033116340637,
"learning_rate": 5e-05,
"loss": 2.086463451385498,
"step": 1272
},
{
"epoch": 0.44423522848126645,
"grad_norm": 0.313363641500473,
"learning_rate": 5e-05,
"loss": 1.9883931875228882,
"step": 1274
},
{
"epoch": 0.4449326150251931,
"grad_norm": 0.26909589767456055,
"learning_rate": 5e-05,
"loss": 2.0722873210906982,
"step": 1276
},
{
"epoch": 0.44563000156911975,
"grad_norm": 0.17645885050296783,
"learning_rate": 5e-05,
"loss": 2.0468873977661133,
"step": 1278
},
{
"epoch": 0.44632738811304634,
"grad_norm": 0.21794313192367554,
"learning_rate": 5e-05,
"loss": 1.936154842376709,
"step": 1280
},
{
"epoch": 0.447024774656973,
"grad_norm": 0.21685075759887695,
"learning_rate": 5e-05,
"loss": 2.084484577178955,
"step": 1282
},
{
"epoch": 0.44772216120089964,
"grad_norm": 0.2649660110473633,
"learning_rate": 5e-05,
"loss": 2.0763959884643555,
"step": 1284
},
{
"epoch": 0.4484195477448263,
"grad_norm": 0.2876056134700775,
"learning_rate": 5e-05,
"loss": 2.029877185821533,
"step": 1286
},
{
"epoch": 0.4491169342887529,
"grad_norm": 0.2842501103878021,
"learning_rate": 5e-05,
"loss": 1.8488447666168213,
"step": 1288
},
{
"epoch": 0.4498143208326795,
"grad_norm": 0.31609490513801575,
"learning_rate": 5e-05,
"loss": 1.6472361087799072,
"step": 1290
},
{
"epoch": 0.4505117073766062,
"grad_norm": 0.18136271834373474,
"learning_rate": 5e-05,
"loss": 1.9367755651474,
"step": 1292
},
{
"epoch": 0.4512090939205328,
"grad_norm": 0.18368104100227356,
"learning_rate": 5e-05,
"loss": 2.182744026184082,
"step": 1294
},
{
"epoch": 0.4519064804644594,
"grad_norm": 0.2922597825527191,
"learning_rate": 5e-05,
"loss": 2.1172962188720703,
"step": 1296
},
{
"epoch": 0.45260386700838606,
"grad_norm": 0.18549898266792297,
"learning_rate": 5e-05,
"loss": 1.8879756927490234,
"step": 1298
},
{
"epoch": 0.4533012535523127,
"grad_norm": 0.2708760201931,
"learning_rate": 5e-05,
"loss": 1.9968003034591675,
"step": 1300
},
{
"epoch": 0.45399864009623936,
"grad_norm": 0.18067607283592224,
"learning_rate": 5e-05,
"loss": 1.9076361656188965,
"step": 1302
},
{
"epoch": 0.45469602664016595,
"grad_norm": 0.17177116870880127,
"learning_rate": 5e-05,
"loss": 2.058980941772461,
"step": 1304
},
{
"epoch": 0.4553934131840926,
"grad_norm": 0.21306143701076508,
"learning_rate": 5e-05,
"loss": 2.1413674354553223,
"step": 1306
},
{
"epoch": 0.45609079972801925,
"grad_norm": 0.2684168815612793,
"learning_rate": 5e-05,
"loss": 1.8674421310424805,
"step": 1308
},
{
"epoch": 0.4567881862719459,
"grad_norm": 0.1928548961877823,
"learning_rate": 5e-05,
"loss": 2.109370231628418,
"step": 1310
},
{
"epoch": 0.45748557281587254,
"grad_norm": 0.22075115144252777,
"learning_rate": 5e-05,
"loss": 2.0332212448120117,
"step": 1312
},
{
"epoch": 0.45818295935979914,
"grad_norm": 0.22620245814323425,
"learning_rate": 5e-05,
"loss": 1.9118460416793823,
"step": 1314
},
{
"epoch": 0.4588803459037258,
"grad_norm": 0.17279599606990814,
"learning_rate": 5e-05,
"loss": 1.975062370300293,
"step": 1316
},
{
"epoch": 0.45957773244765243,
"grad_norm": 0.22477290034294128,
"learning_rate": 5e-05,
"loss": 2.146522045135498,
"step": 1318
},
{
"epoch": 0.4602751189915791,
"grad_norm": 0.30667194724082947,
"learning_rate": 5e-05,
"loss": 1.933511734008789,
"step": 1320
},
{
"epoch": 0.46097250553550567,
"grad_norm": 0.21251042187213898,
"learning_rate": 5e-05,
"loss": 2.0979998111724854,
"step": 1322
},
{
"epoch": 0.4616698920794323,
"grad_norm": 0.21532893180847168,
"learning_rate": 5e-05,
"loss": 1.8038239479064941,
"step": 1324
},
{
"epoch": 0.46236727862335897,
"grad_norm": 0.2114000916481018,
"learning_rate": 5e-05,
"loss": 2.008584499359131,
"step": 1326
},
{
"epoch": 0.4630646651672856,
"grad_norm": 0.16026270389556885,
"learning_rate": 5e-05,
"loss": 2.036628246307373,
"step": 1328
},
{
"epoch": 0.4637620517112122,
"grad_norm": 0.2072455734014511,
"learning_rate": 5e-05,
"loss": 1.9640253782272339,
"step": 1330
},
{
"epoch": 0.46445943825513886,
"grad_norm": 0.2008819282054901,
"learning_rate": 5e-05,
"loss": 2.1391754150390625,
"step": 1332
},
{
"epoch": 0.4651568247990655,
"grad_norm": 0.15363521873950958,
"learning_rate": 5e-05,
"loss": 2.21854305267334,
"step": 1334
},
{
"epoch": 0.46585421134299215,
"grad_norm": 0.30436429381370544,
"learning_rate": 5e-05,
"loss": 2.0346920490264893,
"step": 1336
},
{
"epoch": 0.46655159788691875,
"grad_norm": 0.20182740688323975,
"learning_rate": 5e-05,
"loss": 1.9244691133499146,
"step": 1338
},
{
"epoch": 0.4672489844308454,
"grad_norm": 0.27045613527297974,
"learning_rate": 5e-05,
"loss": 2.037353038787842,
"step": 1340
},
{
"epoch": 0.46794637097477204,
"grad_norm": 0.1664476990699768,
"learning_rate": 5e-05,
"loss": 2.0294156074523926,
"step": 1342
},
{
"epoch": 0.4686437575186987,
"grad_norm": 0.20976664125919342,
"learning_rate": 5e-05,
"loss": 1.9494667053222656,
"step": 1344
},
{
"epoch": 0.46934114406262534,
"grad_norm": 0.22498564422130585,
"learning_rate": 5e-05,
"loss": 1.9336860179901123,
"step": 1346
},
{
"epoch": 0.47003853060655193,
"grad_norm": 0.17451924085617065,
"learning_rate": 5e-05,
"loss": 2.027860641479492,
"step": 1348
},
{
"epoch": 0.4707359171504786,
"grad_norm": 0.28990885615348816,
"learning_rate": 5e-05,
"loss": 1.8952656984329224,
"step": 1350
},
{
"epoch": 0.4714333036944052,
"grad_norm": 0.2333941012620926,
"learning_rate": 5e-05,
"loss": 1.7821979522705078,
"step": 1352
},
{
"epoch": 0.4721306902383319,
"grad_norm": 0.231252983212471,
"learning_rate": 5e-05,
"loss": 1.9425376653671265,
"step": 1354
},
{
"epoch": 0.47282807678225847,
"grad_norm": 0.21064279973506927,
"learning_rate": 5e-05,
"loss": 1.7968411445617676,
"step": 1356
},
{
"epoch": 0.4735254633261851,
"grad_norm": 0.2252817004919052,
"learning_rate": 5e-05,
"loss": 1.9492859840393066,
"step": 1358
},
{
"epoch": 0.47422284987011176,
"grad_norm": 0.2374482899904251,
"learning_rate": 5e-05,
"loss": 2.0691514015197754,
"step": 1360
},
{
"epoch": 0.4749202364140384,
"grad_norm": 0.16628314554691315,
"learning_rate": 5e-05,
"loss": 1.9889640808105469,
"step": 1362
},
{
"epoch": 0.475617622957965,
"grad_norm": 0.23188260197639465,
"learning_rate": 5e-05,
"loss": 1.826238989830017,
"step": 1364
},
{
"epoch": 0.47631500950189165,
"grad_norm": 0.23864124715328217,
"learning_rate": 5e-05,
"loss": 1.8316365480422974,
"step": 1366
},
{
"epoch": 0.4770123960458183,
"grad_norm": 0.17230239510536194,
"learning_rate": 5e-05,
"loss": 2.108180522918701,
"step": 1368
},
{
"epoch": 0.47770978258974495,
"grad_norm": 0.3431381583213806,
"learning_rate": 5e-05,
"loss": 1.927293062210083,
"step": 1370
},
{
"epoch": 0.47840716913367154,
"grad_norm": 0.1808742880821228,
"learning_rate": 5e-05,
"loss": 2.0206215381622314,
"step": 1372
},
{
"epoch": 0.4791045556775982,
"grad_norm": 0.19927719235420227,
"learning_rate": 5e-05,
"loss": 2.071760654449463,
"step": 1374
},
{
"epoch": 0.47980194222152484,
"grad_norm": 0.23531778156757355,
"learning_rate": 5e-05,
"loss": 2.0391430854797363,
"step": 1376
},
{
"epoch": 0.4804993287654515,
"grad_norm": 0.1925334483385086,
"learning_rate": 5e-05,
"loss": 1.8715038299560547,
"step": 1378
},
{
"epoch": 0.48119671530937813,
"grad_norm": 0.17263978719711304,
"learning_rate": 5e-05,
"loss": 2.0224063396453857,
"step": 1380
},
{
"epoch": 0.4818941018533047,
"grad_norm": 0.24466554820537567,
"learning_rate": 5e-05,
"loss": 2.037820339202881,
"step": 1382
},
{
"epoch": 0.48259148839723137,
"grad_norm": 0.3562721014022827,
"learning_rate": 5e-05,
"loss": 1.962568998336792,
"step": 1384
},
{
"epoch": 0.483288874941158,
"grad_norm": 0.19220297038555145,
"learning_rate": 5e-05,
"loss": 2.120224952697754,
"step": 1386
},
{
"epoch": 0.48398626148508467,
"grad_norm": 0.17400771379470825,
"learning_rate": 5e-05,
"loss": 1.9022951126098633,
"step": 1388
},
{
"epoch": 0.48468364802901126,
"grad_norm": 0.2036677896976471,
"learning_rate": 5e-05,
"loss": 2.034552574157715,
"step": 1390
},
{
"epoch": 0.4853810345729379,
"grad_norm": 0.23469923436641693,
"learning_rate": 5e-05,
"loss": 2.1075315475463867,
"step": 1392
},
{
"epoch": 0.48607842111686456,
"grad_norm": 0.19877077639102936,
"learning_rate": 5e-05,
"loss": 2.0443668365478516,
"step": 1394
},
{
"epoch": 0.4867758076607912,
"grad_norm": 0.21028253436088562,
"learning_rate": 5e-05,
"loss": 2.05700421333313,
"step": 1396
},
{
"epoch": 0.4874731942047178,
"grad_norm": 0.2600529193878174,
"learning_rate": 5e-05,
"loss": 1.984779715538025,
"step": 1398
},
{
"epoch": 0.48817058074864444,
"grad_norm": 0.2925497591495514,
"learning_rate": 5e-05,
"loss": 1.9900577068328857,
"step": 1400
},
{
"epoch": 0.4888679672925711,
"grad_norm": 0.21559077501296997,
"learning_rate": 5e-05,
"loss": 1.9484243392944336,
"step": 1402
},
{
"epoch": 0.48956535383649774,
"grad_norm": 0.22062966227531433,
"learning_rate": 5e-05,
"loss": 1.7250257730484009,
"step": 1404
},
{
"epoch": 0.49026274038042433,
"grad_norm": 0.2718144357204437,
"learning_rate": 5e-05,
"loss": 2.0089738368988037,
"step": 1406
},
{
"epoch": 0.490960126924351,
"grad_norm": 0.21445487439632416,
"learning_rate": 5e-05,
"loss": 1.8109028339385986,
"step": 1408
},
{
"epoch": 0.49165751346827763,
"grad_norm": 0.28173160552978516,
"learning_rate": 5e-05,
"loss": 1.964066743850708,
"step": 1410
},
{
"epoch": 0.4923549000122043,
"grad_norm": 0.23372000455856323,
"learning_rate": 5e-05,
"loss": 1.9023646116256714,
"step": 1412
},
{
"epoch": 0.4930522865561309,
"grad_norm": 0.20902119576931,
"learning_rate": 5e-05,
"loss": 1.8483302593231201,
"step": 1414
},
{
"epoch": 0.4937496731000575,
"grad_norm": 0.20217423141002655,
"learning_rate": 5e-05,
"loss": 2.0042505264282227,
"step": 1416
},
{
"epoch": 0.49444705964398417,
"grad_norm": 0.1686791032552719,
"learning_rate": 5e-05,
"loss": 2.167377233505249,
"step": 1418
},
{
"epoch": 0.4951444461879108,
"grad_norm": 0.21020229160785675,
"learning_rate": 5e-05,
"loss": 1.88352370262146,
"step": 1420
},
{
"epoch": 0.49584183273183746,
"grad_norm": 0.16124327480793,
"learning_rate": 5e-05,
"loss": 1.8495641946792603,
"step": 1422
},
{
"epoch": 0.49653921927576405,
"grad_norm": 0.19446009397506714,
"learning_rate": 5e-05,
"loss": 1.8798768520355225,
"step": 1424
},
{
"epoch": 0.4972366058196907,
"grad_norm": 0.17021384835243225,
"learning_rate": 5e-05,
"loss": 2.11187744140625,
"step": 1426
},
{
"epoch": 0.49793399236361735,
"grad_norm": 0.20835986733436584,
"learning_rate": 5e-05,
"loss": 1.9984071254730225,
"step": 1428
},
{
"epoch": 0.498631378907544,
"grad_norm": 0.18223054707050323,
"learning_rate": 5e-05,
"loss": 1.721818208694458,
"step": 1430
},
{
"epoch": 0.4993287654514706,
"grad_norm": 0.202471524477005,
"learning_rate": 5e-05,
"loss": 2.026165723800659,
"step": 1432
},
{
"epoch": 0.5000261519953972,
"grad_norm": 0.16094128787517548,
"learning_rate": 5e-05,
"loss": 1.8392627239227295,
"step": 1434
},
{
"epoch": 0.5007235385393238,
"grad_norm": 0.36537352204322815,
"learning_rate": 5e-05,
"loss": 1.818052887916565,
"step": 1436
},
{
"epoch": 0.5014209250832505,
"grad_norm": 0.3065716326236725,
"learning_rate": 5e-05,
"loss": 1.5908174514770508,
"step": 1438
},
{
"epoch": 0.5021183116271771,
"grad_norm": 0.315999835729599,
"learning_rate": 5e-05,
"loss": 2.0305862426757812,
"step": 1440
},
{
"epoch": 0.5028156981711038,
"grad_norm": 0.22419418394565582,
"learning_rate": 5e-05,
"loss": 2.0770304203033447,
"step": 1442
},
{
"epoch": 0.5035130847150304,
"grad_norm": 0.19013993442058563,
"learning_rate": 5e-05,
"loss": 1.9330451488494873,
"step": 1444
},
{
"epoch": 0.504210471258957,
"grad_norm": 0.23034580051898956,
"learning_rate": 5e-05,
"loss": 2.0115108489990234,
"step": 1446
},
{
"epoch": 0.5049078578028837,
"grad_norm": 0.22189804911613464,
"learning_rate": 5e-05,
"loss": 1.9059724807739258,
"step": 1448
},
{
"epoch": 0.5056052443468103,
"grad_norm": 0.24250493943691254,
"learning_rate": 5e-05,
"loss": 1.9216318130493164,
"step": 1450
},
{
"epoch": 0.506302630890737,
"grad_norm": 0.1884554922580719,
"learning_rate": 5e-05,
"loss": 2.0774998664855957,
"step": 1452
},
{
"epoch": 0.5070000174346636,
"grad_norm": 0.32978472113609314,
"learning_rate": 5e-05,
"loss": 1.7872614860534668,
"step": 1454
},
{
"epoch": 0.5076974039785902,
"grad_norm": 0.20328378677368164,
"learning_rate": 5e-05,
"loss": 2.087057590484619,
"step": 1456
},
{
"epoch": 0.5083947905225169,
"grad_norm": 0.1932574063539505,
"learning_rate": 5e-05,
"loss": 1.9663078784942627,
"step": 1458
},
{
"epoch": 0.5090921770664435,
"grad_norm": 0.2333851009607315,
"learning_rate": 5e-05,
"loss": 2.114800453186035,
"step": 1460
},
{
"epoch": 0.5097895636103701,
"grad_norm": 0.2896854877471924,
"learning_rate": 5e-05,
"loss": 1.821686863899231,
"step": 1462
},
{
"epoch": 0.5104869501542968,
"grad_norm": 0.15735310316085815,
"learning_rate": 5e-05,
"loss": 2.0014073848724365,
"step": 1464
},
{
"epoch": 0.5111843366982234,
"grad_norm": 0.20952001214027405,
"learning_rate": 5e-05,
"loss": 1.9321421384811401,
"step": 1466
},
{
"epoch": 0.5118817232421501,
"grad_norm": 0.26672106981277466,
"learning_rate": 5e-05,
"loss": 2.0439276695251465,
"step": 1468
},
{
"epoch": 0.5125791097860767,
"grad_norm": 0.24956052005290985,
"learning_rate": 5e-05,
"loss": 2.0869853496551514,
"step": 1470
},
{
"epoch": 0.5132764963300033,
"grad_norm": 0.1717880517244339,
"learning_rate": 5e-05,
"loss": 1.9398398399353027,
"step": 1472
},
{
"epoch": 0.51397388287393,
"grad_norm": 0.22402872145175934,
"learning_rate": 5e-05,
"loss": 1.7385175228118896,
"step": 1474
},
{
"epoch": 0.5146712694178566,
"grad_norm": 0.22106806933879852,
"learning_rate": 5e-05,
"loss": 2.1921401023864746,
"step": 1476
},
{
"epoch": 0.5153686559617832,
"grad_norm": 0.21225227415561676,
"learning_rate": 5e-05,
"loss": 1.813704490661621,
"step": 1478
},
{
"epoch": 0.5160660425057099,
"grad_norm": 0.24747970700263977,
"learning_rate": 5e-05,
"loss": 2.0386838912963867,
"step": 1480
},
{
"epoch": 0.5167634290496365,
"grad_norm": 0.2071690410375595,
"learning_rate": 5e-05,
"loss": 1.9998681545257568,
"step": 1482
},
{
"epoch": 0.5174608155935632,
"grad_norm": 0.19989708065986633,
"learning_rate": 5e-05,
"loss": 1.9855647087097168,
"step": 1484
},
{
"epoch": 0.5181582021374898,
"grad_norm": 0.37738072872161865,
"learning_rate": 5e-05,
"loss": 1.9317657947540283,
"step": 1486
},
{
"epoch": 0.5188555886814163,
"grad_norm": 0.19475798308849335,
"learning_rate": 5e-05,
"loss": 2.1612281799316406,
"step": 1488
},
{
"epoch": 0.519552975225343,
"grad_norm": 0.16933269798755646,
"learning_rate": 5e-05,
"loss": 1.8891172409057617,
"step": 1490
},
{
"epoch": 0.5202503617692696,
"grad_norm": 0.2092454433441162,
"learning_rate": 5e-05,
"loss": 1.9128150939941406,
"step": 1492
},
{
"epoch": 0.5209477483131963,
"grad_norm": 0.1707553267478943,
"learning_rate": 5e-05,
"loss": 2.026205539703369,
"step": 1494
},
{
"epoch": 0.5216451348571229,
"grad_norm": 0.26486045122146606,
"learning_rate": 5e-05,
"loss": 1.932251214981079,
"step": 1496
},
{
"epoch": 0.5223425214010495,
"grad_norm": 0.24055586755275726,
"learning_rate": 5e-05,
"loss": 2.102813482284546,
"step": 1498
},
{
"epoch": 0.5230399079449762,
"grad_norm": 0.22778025269508362,
"learning_rate": 5e-05,
"loss": 2.025139093399048,
"step": 1500
},
{
"epoch": 0.5237372944889028,
"grad_norm": 0.21262387931346893,
"learning_rate": 5e-05,
"loss": 1.9433798789978027,
"step": 1502
},
{
"epoch": 0.5244346810328294,
"grad_norm": 0.22720636427402496,
"learning_rate": 5e-05,
"loss": 1.8567476272583008,
"step": 1504
},
{
"epoch": 0.5251320675767561,
"grad_norm": 0.21710364520549774,
"learning_rate": 5e-05,
"loss": 1.9706692695617676,
"step": 1506
},
{
"epoch": 0.5258294541206827,
"grad_norm": 0.16479766368865967,
"learning_rate": 5e-05,
"loss": 2.0109052658081055,
"step": 1508
},
{
"epoch": 0.5265268406646094,
"grad_norm": 0.2113187611103058,
"learning_rate": 5e-05,
"loss": 1.8496572971343994,
"step": 1510
},
{
"epoch": 0.527224227208536,
"grad_norm": 0.2933638095855713,
"learning_rate": 5e-05,
"loss": 2.052328586578369,
"step": 1512
},
{
"epoch": 0.5279216137524626,
"grad_norm": 0.2246863842010498,
"learning_rate": 5e-05,
"loss": 1.9460456371307373,
"step": 1514
},
{
"epoch": 0.5286190002963893,
"grad_norm": 0.1951591968536377,
"learning_rate": 5e-05,
"loss": 1.9435811042785645,
"step": 1516
},
{
"epoch": 0.5293163868403159,
"grad_norm": 0.20596501231193542,
"learning_rate": 5e-05,
"loss": 1.9414966106414795,
"step": 1518
},
{
"epoch": 0.5300137733842426,
"grad_norm": 0.21787771582603455,
"learning_rate": 5e-05,
"loss": 1.9627716541290283,
"step": 1520
},
{
"epoch": 0.5307111599281692,
"grad_norm": 0.24987459182739258,
"learning_rate": 5e-05,
"loss": 2.041393280029297,
"step": 1522
},
{
"epoch": 0.5314085464720958,
"grad_norm": 0.22474417090415955,
"learning_rate": 5e-05,
"loss": 2.2523980140686035,
"step": 1524
},
{
"epoch": 0.5321059330160225,
"grad_norm": 0.2154618799686432,
"learning_rate": 5e-05,
"loss": 1.9683117866516113,
"step": 1526
},
{
"epoch": 0.5328033195599491,
"grad_norm": 0.2014053463935852,
"learning_rate": 5e-05,
"loss": 1.9071648120880127,
"step": 1528
},
{
"epoch": 0.5335007061038757,
"grad_norm": 0.2900610864162445,
"learning_rate": 5e-05,
"loss": 2.0379743576049805,
"step": 1530
},
{
"epoch": 0.5341980926478024,
"grad_norm": 0.19507604837417603,
"learning_rate": 5e-05,
"loss": 1.7710990905761719,
"step": 1532
},
{
"epoch": 0.534895479191729,
"grad_norm": 0.25735238194465637,
"learning_rate": 5e-05,
"loss": 1.878333568572998,
"step": 1534
},
{
"epoch": 0.5355928657356557,
"grad_norm": 0.23907451331615448,
"learning_rate": 5e-05,
"loss": 1.968424677848816,
"step": 1536
},
{
"epoch": 0.5362902522795823,
"grad_norm": 0.29978978633880615,
"learning_rate": 5e-05,
"loss": 1.9405322074890137,
"step": 1538
},
{
"epoch": 0.5369876388235089,
"grad_norm": 0.19385389983654022,
"learning_rate": 5e-05,
"loss": 1.9528419971466064,
"step": 1540
},
{
"epoch": 0.5376850253674356,
"grad_norm": 0.23425279557704926,
"learning_rate": 5e-05,
"loss": 2.1229302883148193,
"step": 1542
},
{
"epoch": 0.5383824119113622,
"grad_norm": 0.21549579501152039,
"learning_rate": 5e-05,
"loss": 2.018282890319824,
"step": 1544
},
{
"epoch": 0.5390797984552888,
"grad_norm": 0.28046560287475586,
"learning_rate": 5e-05,
"loss": 1.7114261388778687,
"step": 1546
},
{
"epoch": 0.5397771849992155,
"grad_norm": 0.24298356473445892,
"learning_rate": 5e-05,
"loss": 2.050609588623047,
"step": 1548
},
{
"epoch": 0.540474571543142,
"grad_norm": 0.18314428627490997,
"learning_rate": 5e-05,
"loss": 2.1509413719177246,
"step": 1550
},
{
"epoch": 0.5411719580870687,
"grad_norm": 0.24424991011619568,
"learning_rate": 5e-05,
"loss": 1.9574331045150757,
"step": 1552
},
{
"epoch": 0.5418693446309953,
"grad_norm": 0.20439790189266205,
"learning_rate": 5e-05,
"loss": 1.9486185312271118,
"step": 1554
},
{
"epoch": 0.5425667311749219,
"grad_norm": 0.18379662930965424,
"learning_rate": 5e-05,
"loss": 1.8366260528564453,
"step": 1556
},
{
"epoch": 0.5432641177188486,
"grad_norm": 0.2572096884250641,
"learning_rate": 5e-05,
"loss": 1.8434381484985352,
"step": 1558
},
{
"epoch": 0.5439615042627752,
"grad_norm": 0.2464679777622223,
"learning_rate": 5e-05,
"loss": 1.802624225616455,
"step": 1560
},
{
"epoch": 0.5446588908067019,
"grad_norm": 0.17593611776828766,
"learning_rate": 5e-05,
"loss": 1.893507480621338,
"step": 1562
},
{
"epoch": 0.5453562773506285,
"grad_norm": 0.320984423160553,
"learning_rate": 5e-05,
"loss": 1.7396756410598755,
"step": 1564
},
{
"epoch": 0.5460536638945551,
"grad_norm": 0.23846492171287537,
"learning_rate": 5e-05,
"loss": 1.8965473175048828,
"step": 1566
},
{
"epoch": 0.5467510504384818,
"grad_norm": 0.1506342887878418,
"learning_rate": 5e-05,
"loss": 1.9256426095962524,
"step": 1568
},
{
"epoch": 0.5474484369824084,
"grad_norm": 0.17887084186077118,
"learning_rate": 5e-05,
"loss": 2.072817802429199,
"step": 1570
},
{
"epoch": 0.548145823526335,
"grad_norm": 0.20621606707572937,
"learning_rate": 5e-05,
"loss": 1.6939451694488525,
"step": 1572
},
{
"epoch": 0.5488432100702617,
"grad_norm": 0.27097389101982117,
"learning_rate": 5e-05,
"loss": 1.9146214723587036,
"step": 1574
},
{
"epoch": 0.5495405966141883,
"grad_norm": 0.16874749958515167,
"learning_rate": 5e-05,
"loss": 2.0716683864593506,
"step": 1576
},
{
"epoch": 0.550237983158115,
"grad_norm": 0.200093075633049,
"learning_rate": 5e-05,
"loss": 1.9811136722564697,
"step": 1578
},
{
"epoch": 0.5509353697020416,
"grad_norm": 0.2207585871219635,
"learning_rate": 5e-05,
"loss": 1.954552173614502,
"step": 1580
},
{
"epoch": 0.5516327562459682,
"grad_norm": 0.2329169660806656,
"learning_rate": 5e-05,
"loss": 2.050635576248169,
"step": 1582
},
{
"epoch": 0.5523301427898949,
"grad_norm": 0.21007491648197174,
"learning_rate": 5e-05,
"loss": 1.982131004333496,
"step": 1584
},
{
"epoch": 0.5530275293338215,
"grad_norm": 0.15652252733707428,
"learning_rate": 5e-05,
"loss": 1.9359612464904785,
"step": 1586
},
{
"epoch": 0.5537249158777482,
"grad_norm": 0.21786396205425262,
"learning_rate": 5e-05,
"loss": 1.7455949783325195,
"step": 1588
},
{
"epoch": 0.5544223024216748,
"grad_norm": 0.20270538330078125,
"learning_rate": 5e-05,
"loss": 1.8889861106872559,
"step": 1590
},
{
"epoch": 0.5551196889656014,
"grad_norm": 0.26479393243789673,
"learning_rate": 5e-05,
"loss": 1.8715715408325195,
"step": 1592
},
{
"epoch": 0.5558170755095281,
"grad_norm": 0.25771623849868774,
"learning_rate": 5e-05,
"loss": 1.8482694625854492,
"step": 1594
},
{
"epoch": 0.5565144620534547,
"grad_norm": 0.210859015583992,
"learning_rate": 5e-05,
"loss": 1.9231295585632324,
"step": 1596
},
{
"epoch": 0.5572118485973813,
"grad_norm": 0.2072077989578247,
"learning_rate": 5e-05,
"loss": 1.7906043529510498,
"step": 1598
},
{
"epoch": 0.557909235141308,
"grad_norm": 0.29259616136550903,
"learning_rate": 5e-05,
"loss": 1.9715745449066162,
"step": 1600
},
{
"epoch": 0.5586066216852346,
"grad_norm": 0.24774612486362457,
"learning_rate": 5e-05,
"loss": 2.081681728363037,
"step": 1602
},
{
"epoch": 0.5593040082291613,
"grad_norm": 0.24854712188243866,
"learning_rate": 5e-05,
"loss": 1.833552598953247,
"step": 1604
},
{
"epoch": 0.5600013947730879,
"grad_norm": 0.2399366796016693,
"learning_rate": 5e-05,
"loss": 2.124068021774292,
"step": 1606
},
{
"epoch": 0.5606987813170144,
"grad_norm": 0.18966670334339142,
"learning_rate": 5e-05,
"loss": 1.7892274856567383,
"step": 1608
},
{
"epoch": 0.5613961678609412,
"grad_norm": 0.23364785313606262,
"learning_rate": 5e-05,
"loss": 1.8081241846084595,
"step": 1610
},
{
"epoch": 0.5620935544048677,
"grad_norm": 0.18954408168792725,
"learning_rate": 5e-05,
"loss": 2.034299373626709,
"step": 1612
},
{
"epoch": 0.5627909409487943,
"grad_norm": 0.1632585972547531,
"learning_rate": 5e-05,
"loss": 1.9397029876708984,
"step": 1614
},
{
"epoch": 0.563488327492721,
"grad_norm": 0.2352249175310135,
"learning_rate": 5e-05,
"loss": 1.9451236724853516,
"step": 1616
},
{
"epoch": 0.5641857140366476,
"grad_norm": 0.2975192070007324,
"learning_rate": 5e-05,
"loss": 1.8692302703857422,
"step": 1618
},
{
"epoch": 0.5648831005805743,
"grad_norm": 0.25654807686805725,
"learning_rate": 5e-05,
"loss": 1.9424086809158325,
"step": 1620
},
{
"epoch": 0.5655804871245009,
"grad_norm": 0.2531231641769409,
"learning_rate": 5e-05,
"loss": 2.043811082839966,
"step": 1622
},
{
"epoch": 0.5662778736684275,
"grad_norm": 0.189333975315094,
"learning_rate": 5e-05,
"loss": 1.906083345413208,
"step": 1624
},
{
"epoch": 0.5669752602123542,
"grad_norm": 0.2502206861972809,
"learning_rate": 5e-05,
"loss": 2.0286500453948975,
"step": 1626
},
{
"epoch": 0.5676726467562808,
"grad_norm": 0.19187211990356445,
"learning_rate": 5e-05,
"loss": 1.9105477333068848,
"step": 1628
},
{
"epoch": 0.5683700333002075,
"grad_norm": 0.19181573390960693,
"learning_rate": 5e-05,
"loss": 1.9919695854187012,
"step": 1630
},
{
"epoch": 0.5690674198441341,
"grad_norm": 0.25628000497817993,
"learning_rate": 5e-05,
"loss": 1.9693403244018555,
"step": 1632
},
{
"epoch": 0.5697648063880607,
"grad_norm": 0.29909878969192505,
"learning_rate": 5e-05,
"loss": 1.9884252548217773,
"step": 1634
},
{
"epoch": 0.5704621929319874,
"grad_norm": 0.1668253093957901,
"learning_rate": 5e-05,
"loss": 1.8667051792144775,
"step": 1636
},
{
"epoch": 0.571159579475914,
"grad_norm": 0.3491284251213074,
"learning_rate": 5e-05,
"loss": 1.781627893447876,
"step": 1638
},
{
"epoch": 0.5718569660198406,
"grad_norm": 0.20913377404212952,
"learning_rate": 5e-05,
"loss": 2.139101505279541,
"step": 1640
},
{
"epoch": 0.5725543525637673,
"grad_norm": 0.19669152796268463,
"learning_rate": 5e-05,
"loss": 1.9148974418640137,
"step": 1642
},
{
"epoch": 0.5732517391076939,
"grad_norm": 0.3062213957309723,
"learning_rate": 5e-05,
"loss": 1.8412705659866333,
"step": 1644
},
{
"epoch": 0.5739491256516206,
"grad_norm": 0.2283291071653366,
"learning_rate": 5e-05,
"loss": 1.8515138626098633,
"step": 1646
},
{
"epoch": 0.5746465121955472,
"grad_norm": 0.25086915493011475,
"learning_rate": 5e-05,
"loss": 1.9342883825302124,
"step": 1648
},
{
"epoch": 0.5753438987394738,
"grad_norm": 0.20395423471927643,
"learning_rate": 5e-05,
"loss": 2.053553819656372,
"step": 1650
},
{
"epoch": 0.5760412852834005,
"grad_norm": 0.29043474793434143,
"learning_rate": 5e-05,
"loss": 1.9286761283874512,
"step": 1652
},
{
"epoch": 0.5767386718273271,
"grad_norm": 0.3165014982223511,
"learning_rate": 5e-05,
"loss": 1.993572473526001,
"step": 1654
},
{
"epoch": 0.5774360583712538,
"grad_norm": 0.3387148082256317,
"learning_rate": 5e-05,
"loss": 1.8471314907073975,
"step": 1656
},
{
"epoch": 0.5781334449151804,
"grad_norm": 0.2811840772628784,
"learning_rate": 5e-05,
"loss": 1.8386205434799194,
"step": 1658
},
{
"epoch": 0.578830831459107,
"grad_norm": 0.22342753410339355,
"learning_rate": 5e-05,
"loss": 2.005929470062256,
"step": 1660
},
{
"epoch": 0.5795282180030337,
"grad_norm": 0.25924915075302124,
"learning_rate": 5e-05,
"loss": 1.7976369857788086,
"step": 1662
},
{
"epoch": 0.5802256045469603,
"grad_norm": 0.24489231407642365,
"learning_rate": 5e-05,
"loss": 1.8548369407653809,
"step": 1664
},
{
"epoch": 0.5809229910908869,
"grad_norm": 0.20547856390476227,
"learning_rate": 5e-05,
"loss": 1.8978731632232666,
"step": 1666
},
{
"epoch": 0.5816203776348136,
"grad_norm": 0.4370429515838623,
"learning_rate": 5e-05,
"loss": 1.8970003128051758,
"step": 1668
},
{
"epoch": 0.5823177641787401,
"grad_norm": 0.1849474310874939,
"learning_rate": 5e-05,
"loss": 2.0585944652557373,
"step": 1670
},
{
"epoch": 0.5830151507226669,
"grad_norm": 0.19704462587833405,
"learning_rate": 5e-05,
"loss": 1.732386827468872,
"step": 1672
},
{
"epoch": 0.5837125372665934,
"grad_norm": 0.43086493015289307,
"learning_rate": 5e-05,
"loss": 1.7718791961669922,
"step": 1674
},
{
"epoch": 0.58440992381052,
"grad_norm": 0.20836947858333588,
"learning_rate": 5e-05,
"loss": 2.006824016571045,
"step": 1676
},
{
"epoch": 0.5851073103544467,
"grad_norm": 0.2442089021205902,
"learning_rate": 5e-05,
"loss": 2.077127456665039,
"step": 1678
},
{
"epoch": 0.5858046968983733,
"grad_norm": 0.22404621541500092,
"learning_rate": 5e-05,
"loss": 2.043806552886963,
"step": 1680
},
{
"epoch": 0.5865020834422999,
"grad_norm": 0.21961356699466705,
"learning_rate": 5e-05,
"loss": 1.772026777267456,
"step": 1682
},
{
"epoch": 0.5871994699862266,
"grad_norm": 0.3041599690914154,
"learning_rate": 5e-05,
"loss": 1.7471873760223389,
"step": 1684
},
{
"epoch": 0.5878968565301532,
"grad_norm": 0.2028559297323227,
"learning_rate": 5e-05,
"loss": 1.827317476272583,
"step": 1686
},
{
"epoch": 0.5885942430740799,
"grad_norm": 0.20505206286907196,
"learning_rate": 5e-05,
"loss": 1.9476141929626465,
"step": 1688
},
{
"epoch": 0.5892916296180065,
"grad_norm": 0.23087021708488464,
"learning_rate": 5e-05,
"loss": 2.1552538871765137,
"step": 1690
},
{
"epoch": 0.5899890161619331,
"grad_norm": 0.15803898870944977,
"learning_rate": 5e-05,
"loss": 2.1053948402404785,
"step": 1692
},
{
"epoch": 0.5906864027058598,
"grad_norm": 0.21020828187465668,
"learning_rate": 5e-05,
"loss": 2.0307366847991943,
"step": 1694
},
{
"epoch": 0.5913837892497864,
"grad_norm": 0.1724986732006073,
"learning_rate": 5e-05,
"loss": 2.0489397048950195,
"step": 1696
},
{
"epoch": 0.5920811757937131,
"grad_norm": 0.19490469992160797,
"learning_rate": 5e-05,
"loss": 1.923473834991455,
"step": 1698
},
{
"epoch": 0.5927785623376397,
"grad_norm": 0.21006463468074799,
"learning_rate": 5e-05,
"loss": 2.0133984088897705,
"step": 1700
},
{
"epoch": 0.5934759488815663,
"grad_norm": 0.3003818094730377,
"learning_rate": 5e-05,
"loss": 1.8472903966903687,
"step": 1702
},
{
"epoch": 0.594173335425493,
"grad_norm": 0.1842869371175766,
"learning_rate": 5e-05,
"loss": 1.8047981262207031,
"step": 1704
},
{
"epoch": 0.5948707219694196,
"grad_norm": 0.2461615949869156,
"learning_rate": 5e-05,
"loss": 2.078711986541748,
"step": 1706
},
{
"epoch": 0.5955681085133462,
"grad_norm": 0.2864431142807007,
"learning_rate": 5e-05,
"loss": 1.946295976638794,
"step": 1708
},
{
"epoch": 0.5962654950572729,
"grad_norm": 0.2841217517852783,
"learning_rate": 5e-05,
"loss": 1.9858349561691284,
"step": 1710
},
{
"epoch": 0.5969628816011995,
"grad_norm": 0.19433815777301788,
"learning_rate": 5e-05,
"loss": 1.826880931854248,
"step": 1712
},
{
"epoch": 0.5976602681451262,
"grad_norm": 0.22107480466365814,
"learning_rate": 5e-05,
"loss": 1.8655105829238892,
"step": 1714
},
{
"epoch": 0.5983576546890528,
"grad_norm": 0.28393471240997314,
"learning_rate": 5e-05,
"loss": 1.908822774887085,
"step": 1716
},
{
"epoch": 0.5990550412329794,
"grad_norm": 0.2021571695804596,
"learning_rate": 5e-05,
"loss": 1.929540753364563,
"step": 1718
},
{
"epoch": 0.5997524277769061,
"grad_norm": 0.23796068131923676,
"learning_rate": 5e-05,
"loss": 2.007427930831909,
"step": 1720
},
{
"epoch": 0.6004498143208327,
"grad_norm": 0.24815872311592102,
"learning_rate": 5e-05,
"loss": 1.9182190895080566,
"step": 1722
},
{
"epoch": 0.6011472008647594,
"grad_norm": 0.1572408676147461,
"learning_rate": 5e-05,
"loss": 1.959277868270874,
"step": 1724
},
{
"epoch": 0.601844587408686,
"grad_norm": 0.21377843618392944,
"learning_rate": 5e-05,
"loss": 2.175769567489624,
"step": 1726
},
{
"epoch": 0.6025419739526126,
"grad_norm": 0.2010141909122467,
"learning_rate": 5e-05,
"loss": 2.1774816513061523,
"step": 1728
},
{
"epoch": 0.6032393604965393,
"grad_norm": 0.2195289582014084,
"learning_rate": 5e-05,
"loss": 1.7632919549942017,
"step": 1730
},
{
"epoch": 0.6039367470404658,
"grad_norm": 0.21566610038280487,
"learning_rate": 5e-05,
"loss": 1.9510669708251953,
"step": 1732
},
{
"epoch": 0.6046341335843924,
"grad_norm": 0.18458028137683868,
"learning_rate": 5e-05,
"loss": 2.1417834758758545,
"step": 1734
},
{
"epoch": 0.6053315201283191,
"grad_norm": 0.27364224195480347,
"learning_rate": 5e-05,
"loss": 2.0847959518432617,
"step": 1736
},
{
"epoch": 0.6060289066722457,
"grad_norm": 0.20937108993530273,
"learning_rate": 5e-05,
"loss": 2.1932859420776367,
"step": 1738
},
{
"epoch": 0.6067262932161724,
"grad_norm": 0.24981026351451874,
"learning_rate": 5e-05,
"loss": 1.7620201110839844,
"step": 1740
},
{
"epoch": 0.607423679760099,
"grad_norm": 0.22395093739032745,
"learning_rate": 5e-05,
"loss": 1.9993212223052979,
"step": 1742
},
{
"epoch": 0.6081210663040256,
"grad_norm": 0.27337807416915894,
"learning_rate": 5e-05,
"loss": 2.0806846618652344,
"step": 1744
},
{
"epoch": 0.6088184528479523,
"grad_norm": 0.226410910487175,
"learning_rate": 5e-05,
"loss": 1.8875980377197266,
"step": 1746
},
{
"epoch": 0.6095158393918789,
"grad_norm": 0.22300706803798676,
"learning_rate": 5e-05,
"loss": 1.537036418914795,
"step": 1748
},
{
"epoch": 0.6102132259358055,
"grad_norm": 0.1922428160905838,
"learning_rate": 5e-05,
"loss": 1.8651649951934814,
"step": 1750
},
{
"epoch": 0.6109106124797322,
"grad_norm": 0.15310291945934296,
"learning_rate": 5e-05,
"loss": 1.9449410438537598,
"step": 1752
},
{
"epoch": 0.6116079990236588,
"grad_norm": 0.2211032658815384,
"learning_rate": 5e-05,
"loss": 2.1581153869628906,
"step": 1754
},
{
"epoch": 0.6123053855675855,
"grad_norm": 0.2677195966243744,
"learning_rate": 5e-05,
"loss": 1.9254250526428223,
"step": 1756
},
{
"epoch": 0.6130027721115121,
"grad_norm": 0.18960987031459808,
"learning_rate": 5e-05,
"loss": 1.7236549854278564,
"step": 1758
},
{
"epoch": 0.6137001586554387,
"grad_norm": 0.31510594487190247,
"learning_rate": 5e-05,
"loss": 1.95664644241333,
"step": 1760
},
{
"epoch": 0.6143975451993654,
"grad_norm": 0.21830137073993683,
"learning_rate": 5e-05,
"loss": 2.088205575942993,
"step": 1762
},
{
"epoch": 0.615094931743292,
"grad_norm": 0.23890633881092072,
"learning_rate": 5e-05,
"loss": 1.9173531532287598,
"step": 1764
},
{
"epoch": 0.6157923182872187,
"grad_norm": 0.24494515359401703,
"learning_rate": 5e-05,
"loss": 1.929988145828247,
"step": 1766
},
{
"epoch": 0.6164897048311453,
"grad_norm": 0.21958190202713013,
"learning_rate": 5e-05,
"loss": 1.8340654373168945,
"step": 1768
},
{
"epoch": 0.6171870913750719,
"grad_norm": 0.20597727596759796,
"learning_rate": 5e-05,
"loss": 2.064466953277588,
"step": 1770
},
{
"epoch": 0.6178844779189986,
"grad_norm": 0.21612513065338135,
"learning_rate": 5e-05,
"loss": 1.9843177795410156,
"step": 1772
},
{
"epoch": 0.6185818644629252,
"grad_norm": 0.24421793222427368,
"learning_rate": 5e-05,
"loss": 1.828315019607544,
"step": 1774
},
{
"epoch": 0.6192792510068518,
"grad_norm": 0.2955434322357178,
"learning_rate": 5e-05,
"loss": 1.9416334629058838,
"step": 1776
},
{
"epoch": 0.6199766375507785,
"grad_norm": 0.3074122667312622,
"learning_rate": 5e-05,
"loss": 1.8622162342071533,
"step": 1778
},
{
"epoch": 0.6206740240947051,
"grad_norm": 0.2642528712749481,
"learning_rate": 5e-05,
"loss": 1.8647794723510742,
"step": 1780
},
{
"epoch": 0.6213714106386318,
"grad_norm": 0.2782607674598694,
"learning_rate": 5e-05,
"loss": 1.934777021408081,
"step": 1782
},
{
"epoch": 0.6220687971825584,
"grad_norm": 0.16562847793102264,
"learning_rate": 5e-05,
"loss": 2.0389959812164307,
"step": 1784
},
{
"epoch": 0.622766183726485,
"grad_norm": 0.18666969239711761,
"learning_rate": 5e-05,
"loss": 1.9525014162063599,
"step": 1786
},
{
"epoch": 0.6234635702704117,
"grad_norm": 0.19193722307682037,
"learning_rate": 5e-05,
"loss": 1.802701711654663,
"step": 1788
},
{
"epoch": 0.6241609568143383,
"grad_norm": 0.1893835961818695,
"learning_rate": 5e-05,
"loss": 1.9322786331176758,
"step": 1790
},
{
"epoch": 0.624858343358265,
"grad_norm": 0.15804891288280487,
"learning_rate": 5e-05,
"loss": 2.237478256225586,
"step": 1792
},
{
"epoch": 0.6255557299021915,
"grad_norm": 0.28472012281417847,
"learning_rate": 5e-05,
"loss": 1.8306212425231934,
"step": 1794
},
{
"epoch": 0.6262531164461181,
"grad_norm": 0.18115119636058807,
"learning_rate": 5e-05,
"loss": 2.118734121322632,
"step": 1796
},
{
"epoch": 0.6269505029900448,
"grad_norm": 0.17225264012813568,
"learning_rate": 5e-05,
"loss": 2.08634090423584,
"step": 1798
},
{
"epoch": 0.6276478895339714,
"grad_norm": 0.19024088978767395,
"learning_rate": 5e-05,
"loss": 1.741422176361084,
"step": 1800
},
{
"epoch": 0.628345276077898,
"grad_norm": 0.19031323492527008,
"learning_rate": 5e-05,
"loss": 1.8494441509246826,
"step": 1802
},
{
"epoch": 0.6290426626218247,
"grad_norm": 0.21968109905719757,
"learning_rate": 5e-05,
"loss": 1.9487123489379883,
"step": 1804
},
{
"epoch": 0.6297400491657513,
"grad_norm": 0.2162948101758957,
"learning_rate": 5e-05,
"loss": 1.8762587308883667,
"step": 1806
},
{
"epoch": 0.630437435709678,
"grad_norm": 0.1916320025920868,
"learning_rate": 5e-05,
"loss": 2.0723886489868164,
"step": 1808
},
{
"epoch": 0.6311348222536046,
"grad_norm": 0.22107599675655365,
"learning_rate": 5e-05,
"loss": 1.8572006225585938,
"step": 1810
},
{
"epoch": 0.6318322087975312,
"grad_norm": 0.17048195004463196,
"learning_rate": 5e-05,
"loss": 2.0992026329040527,
"step": 1812
},
{
"epoch": 0.6325295953414579,
"grad_norm": 0.1760718822479248,
"learning_rate": 5e-05,
"loss": 2.106672763824463,
"step": 1814
},
{
"epoch": 0.6332269818853845,
"grad_norm": 0.28058508038520813,
"learning_rate": 5e-05,
"loss": 1.952498197555542,
"step": 1816
},
{
"epoch": 0.6339243684293111,
"grad_norm": 0.20093607902526855,
"learning_rate": 5e-05,
"loss": 1.8791782855987549,
"step": 1818
},
{
"epoch": 0.6346217549732378,
"grad_norm": 0.18365196883678436,
"learning_rate": 5e-05,
"loss": 1.9621028900146484,
"step": 1820
},
{
"epoch": 0.6353191415171644,
"grad_norm": 0.22643136978149414,
"learning_rate": 5e-05,
"loss": 2.0192811489105225,
"step": 1822
},
{
"epoch": 0.6360165280610911,
"grad_norm": 0.24730263650417328,
"learning_rate": 5e-05,
"loss": 1.8804388046264648,
"step": 1824
},
{
"epoch": 0.6367139146050177,
"grad_norm": 0.22289031744003296,
"learning_rate": 5e-05,
"loss": 1.768578290939331,
"step": 1826
},
{
"epoch": 0.6374113011489443,
"grad_norm": 0.2560635209083557,
"learning_rate": 5e-05,
"loss": 1.7840062379837036,
"step": 1828
},
{
"epoch": 0.638108687692871,
"grad_norm": 0.17455683648586273,
"learning_rate": 5e-05,
"loss": 1.9665637016296387,
"step": 1830
},
{
"epoch": 0.6388060742367976,
"grad_norm": 0.26289793848991394,
"learning_rate": 5e-05,
"loss": 2.072192668914795,
"step": 1832
},
{
"epoch": 0.6395034607807243,
"grad_norm": 0.1808442771434784,
"learning_rate": 5e-05,
"loss": 1.9312012195587158,
"step": 1834
},
{
"epoch": 0.6402008473246509,
"grad_norm": 0.21631719172000885,
"learning_rate": 5e-05,
"loss": 1.7907676696777344,
"step": 1836
},
{
"epoch": 0.6408982338685775,
"grad_norm": 0.22999930381774902,
"learning_rate": 5e-05,
"loss": 1.7535853385925293,
"step": 1838
},
{
"epoch": 0.6415956204125042,
"grad_norm": 0.2047368735074997,
"learning_rate": 5e-05,
"loss": 1.8089340925216675,
"step": 1840
},
{
"epoch": 0.6422930069564308,
"grad_norm": 0.1811462640762329,
"learning_rate": 5e-05,
"loss": 1.8642290830612183,
"step": 1842
},
{
"epoch": 0.6429903935003574,
"grad_norm": 0.24953307211399078,
"learning_rate": 5e-05,
"loss": 2.032365083694458,
"step": 1844
},
{
"epoch": 0.6436877800442841,
"grad_norm": 0.22790861129760742,
"learning_rate": 5e-05,
"loss": 1.9307947158813477,
"step": 1846
},
{
"epoch": 0.6443851665882107,
"grad_norm": 0.19640816748142242,
"learning_rate": 5e-05,
"loss": 1.737766981124878,
"step": 1848
},
{
"epoch": 0.6450825531321374,
"grad_norm": 0.1855265349149704,
"learning_rate": 5e-05,
"loss": 1.901891827583313,
"step": 1850
},
{
"epoch": 0.645779939676064,
"grad_norm": 0.1998153030872345,
"learning_rate": 5e-05,
"loss": 2.0288705825805664,
"step": 1852
},
{
"epoch": 0.6464773262199905,
"grad_norm": 0.2204006016254425,
"learning_rate": 5e-05,
"loss": 2.0266761779785156,
"step": 1854
},
{
"epoch": 0.6471747127639172,
"grad_norm": 0.29964593052864075,
"learning_rate": 5e-05,
"loss": 1.8914623260498047,
"step": 1856
},
{
"epoch": 0.6478720993078438,
"grad_norm": 0.217330202460289,
"learning_rate": 5e-05,
"loss": 1.7957367897033691,
"step": 1858
},
{
"epoch": 0.6485694858517705,
"grad_norm": 0.26452746987342834,
"learning_rate": 5e-05,
"loss": 1.915511131286621,
"step": 1860
},
{
"epoch": 0.6492668723956971,
"grad_norm": 0.2192654311656952,
"learning_rate": 5e-05,
"loss": 1.7861742973327637,
"step": 1862
},
{
"epoch": 0.6499642589396237,
"grad_norm": 0.27159371972084045,
"learning_rate": 5e-05,
"loss": 1.8134348392486572,
"step": 1864
},
{
"epoch": 0.6506616454835504,
"grad_norm": 0.20807760953903198,
"learning_rate": 5e-05,
"loss": 2.0074868202209473,
"step": 1866
},
{
"epoch": 0.651359032027477,
"grad_norm": 0.24914826452732086,
"learning_rate": 5e-05,
"loss": 1.935471534729004,
"step": 1868
},
{
"epoch": 0.6520564185714036,
"grad_norm": 0.18474753201007843,
"learning_rate": 5e-05,
"loss": 1.9731979370117188,
"step": 1870
},
{
"epoch": 0.6527538051153303,
"grad_norm": 0.1931002289056778,
"learning_rate": 5e-05,
"loss": 2.050971508026123,
"step": 1872
},
{
"epoch": 0.6534511916592569,
"grad_norm": 0.19344626367092133,
"learning_rate": 5e-05,
"loss": 2.109586238861084,
"step": 1874
},
{
"epoch": 0.6541485782031836,
"grad_norm": 0.19569675624370575,
"learning_rate": 5e-05,
"loss": 2.15287184715271,
"step": 1876
},
{
"epoch": 0.6548459647471102,
"grad_norm": 0.24006061255931854,
"learning_rate": 5e-05,
"loss": 1.9154243469238281,
"step": 1878
},
{
"epoch": 0.6555433512910368,
"grad_norm": 0.2992076575756073,
"learning_rate": 5e-05,
"loss": 2.059971809387207,
"step": 1880
},
{
"epoch": 0.6562407378349635,
"grad_norm": 0.1879195272922516,
"learning_rate": 5e-05,
"loss": 1.972310185432434,
"step": 1882
},
{
"epoch": 0.6569381243788901,
"grad_norm": 0.1923721581697464,
"learning_rate": 5e-05,
"loss": 1.8213950395584106,
"step": 1884
},
{
"epoch": 0.6576355109228167,
"grad_norm": 0.31093233823776245,
"learning_rate": 5e-05,
"loss": 1.8817036151885986,
"step": 1886
},
{
"epoch": 0.6583328974667434,
"grad_norm": 0.285758376121521,
"learning_rate": 5e-05,
"loss": 2.0484776496887207,
"step": 1888
},
{
"epoch": 0.65903028401067,
"grad_norm": 0.22534774243831635,
"learning_rate": 5e-05,
"loss": 1.7124214172363281,
"step": 1890
},
{
"epoch": 0.6597276705545967,
"grad_norm": 0.2966947853565216,
"learning_rate": 5e-05,
"loss": 1.831843376159668,
"step": 1892
},
{
"epoch": 0.6604250570985233,
"grad_norm": 0.20267023146152496,
"learning_rate": 5e-05,
"loss": 1.9030438661575317,
"step": 1894
},
{
"epoch": 0.6611224436424499,
"grad_norm": 0.26224076747894287,
"learning_rate": 5e-05,
"loss": 1.999426245689392,
"step": 1896
},
{
"epoch": 0.6618198301863766,
"grad_norm": 0.3326214551925659,
"learning_rate": 5e-05,
"loss": 1.6982316970825195,
"step": 1898
},
{
"epoch": 0.6625172167303032,
"grad_norm": 0.18656374514102936,
"learning_rate": 5e-05,
"loss": 2.2021312713623047,
"step": 1900
},
{
"epoch": 0.6632146032742299,
"grad_norm": 0.21671637892723083,
"learning_rate": 5e-05,
"loss": 1.9708503484725952,
"step": 1902
},
{
"epoch": 0.6639119898181565,
"grad_norm": 0.18599572777748108,
"learning_rate": 5e-05,
"loss": 2.0492098331451416,
"step": 1904
},
{
"epoch": 0.6646093763620831,
"grad_norm": 0.22860100865364075,
"learning_rate": 5e-05,
"loss": 1.881263017654419,
"step": 1906
},
{
"epoch": 0.6653067629060098,
"grad_norm": 0.22557629644870758,
"learning_rate": 5e-05,
"loss": 1.8670588731765747,
"step": 1908
},
{
"epoch": 0.6660041494499364,
"grad_norm": 0.1576979160308838,
"learning_rate": 5e-05,
"loss": 2.156865119934082,
"step": 1910
},
{
"epoch": 0.666701535993863,
"grad_norm": 0.20873196423053741,
"learning_rate": 5e-05,
"loss": 2.1637001037597656,
"step": 1912
},
{
"epoch": 0.6673989225377897,
"grad_norm": 0.21369338035583496,
"learning_rate": 5e-05,
"loss": 2.092163324356079,
"step": 1914
},
{
"epoch": 0.6680963090817162,
"grad_norm": 0.2540096044540405,
"learning_rate": 5e-05,
"loss": 1.9079657793045044,
"step": 1916
},
{
"epoch": 0.668793695625643,
"grad_norm": 0.19144493341445923,
"learning_rate": 5e-05,
"loss": 1.9168627262115479,
"step": 1918
},
{
"epoch": 0.6694910821695695,
"grad_norm": 0.20210625231266022,
"learning_rate": 5e-05,
"loss": 1.975937843322754,
"step": 1920
},
{
"epoch": 0.6701884687134961,
"grad_norm": 0.1825406700372696,
"learning_rate": 5e-05,
"loss": 1.977008581161499,
"step": 1922
},
{
"epoch": 0.6708858552574228,
"grad_norm": 0.17887762188911438,
"learning_rate": 5e-05,
"loss": 1.9100210666656494,
"step": 1924
},
{
"epoch": 0.6715832418013494,
"grad_norm": 0.23779994249343872,
"learning_rate": 5e-05,
"loss": 2.093499183654785,
"step": 1926
},
{
"epoch": 0.6722806283452761,
"grad_norm": 0.1986411213874817,
"learning_rate": 5e-05,
"loss": 1.8822386264801025,
"step": 1928
},
{
"epoch": 0.6729780148892027,
"grad_norm": 0.24904435873031616,
"learning_rate": 5e-05,
"loss": 2.098172903060913,
"step": 1930
},
{
"epoch": 0.6736754014331293,
"grad_norm": 0.220512256026268,
"learning_rate": 5e-05,
"loss": 1.8210785388946533,
"step": 1932
},
{
"epoch": 0.674372787977056,
"grad_norm": 0.202888622879982,
"learning_rate": 5e-05,
"loss": 1.8291442394256592,
"step": 1934
},
{
"epoch": 0.6750701745209826,
"grad_norm": 0.17567485570907593,
"learning_rate": 5e-05,
"loss": 1.904573917388916,
"step": 1936
},
{
"epoch": 0.6757675610649092,
"grad_norm": 0.1913834512233734,
"learning_rate": 5e-05,
"loss": 1.8033437728881836,
"step": 1938
},
{
"epoch": 0.6764649476088359,
"grad_norm": 0.25764867663383484,
"learning_rate": 5e-05,
"loss": 1.8059159517288208,
"step": 1940
},
{
"epoch": 0.6771623341527625,
"grad_norm": 0.19583149254322052,
"learning_rate": 5e-05,
"loss": 1.9165745973587036,
"step": 1942
},
{
"epoch": 0.6778597206966892,
"grad_norm": 0.23066437244415283,
"learning_rate": 5e-05,
"loss": 1.9404631853103638,
"step": 1944
},
{
"epoch": 0.6785571072406158,
"grad_norm": 0.13939902186393738,
"learning_rate": 5e-05,
"loss": 1.7981820106506348,
"step": 1946
},
{
"epoch": 0.6792544937845424,
"grad_norm": 0.20038597285747528,
"learning_rate": 5e-05,
"loss": 2.0496368408203125,
"step": 1948
},
{
"epoch": 0.6799518803284691,
"grad_norm": 0.21413299441337585,
"learning_rate": 5e-05,
"loss": 2.1031856536865234,
"step": 1950
},
{
"epoch": 0.6806492668723957,
"grad_norm": 0.204486683011055,
"learning_rate": 5e-05,
"loss": 1.8733688592910767,
"step": 1952
},
{
"epoch": 0.6813466534163223,
"grad_norm": 0.21280021965503693,
"learning_rate": 5e-05,
"loss": 1.9597742557525635,
"step": 1954
},
{
"epoch": 0.682044039960249,
"grad_norm": 0.2817871570587158,
"learning_rate": 5e-05,
"loss": 1.8568202257156372,
"step": 1956
},
{
"epoch": 0.6827414265041756,
"grad_norm": 0.2650154232978821,
"learning_rate": 5e-05,
"loss": 1.953235387802124,
"step": 1958
},
{
"epoch": 0.6834388130481023,
"grad_norm": 0.22025105357170105,
"learning_rate": 5e-05,
"loss": 1.9382929801940918,
"step": 1960
},
{
"epoch": 0.6841361995920289,
"grad_norm": 0.2663877606391907,
"learning_rate": 5e-05,
"loss": 2.135603189468384,
"step": 1962
},
{
"epoch": 0.6848335861359555,
"grad_norm": 0.17246317863464355,
"learning_rate": 5e-05,
"loss": 2.0259571075439453,
"step": 1964
},
{
"epoch": 0.6855309726798822,
"grad_norm": 0.16565921902656555,
"learning_rate": 5e-05,
"loss": 2.1059682369232178,
"step": 1966
},
{
"epoch": 0.6862283592238088,
"grad_norm": 0.2343938946723938,
"learning_rate": 5e-05,
"loss": 2.0088436603546143,
"step": 1968
},
{
"epoch": 0.6869257457677355,
"grad_norm": 0.19023862481117249,
"learning_rate": 5e-05,
"loss": 1.8418056964874268,
"step": 1970
},
{
"epoch": 0.687623132311662,
"grad_norm": 0.17515328526496887,
"learning_rate": 5e-05,
"loss": 1.9040826559066772,
"step": 1972
},
{
"epoch": 0.6883205188555886,
"grad_norm": 0.26450735330581665,
"learning_rate": 5e-05,
"loss": 1.8515355587005615,
"step": 1974
},
{
"epoch": 0.6890179053995154,
"grad_norm": 0.24569930136203766,
"learning_rate": 5e-05,
"loss": 1.948767900466919,
"step": 1976
},
{
"epoch": 0.6897152919434419,
"grad_norm": 0.2716074585914612,
"learning_rate": 5e-05,
"loss": 1.9467777013778687,
"step": 1978
},
{
"epoch": 0.6904126784873685,
"grad_norm": 0.23883529007434845,
"learning_rate": 5e-05,
"loss": 1.912539005279541,
"step": 1980
},
{
"epoch": 0.6911100650312952,
"grad_norm": 0.18049849569797516,
"learning_rate": 5e-05,
"loss": 1.9098021984100342,
"step": 1982
},
{
"epoch": 0.6918074515752218,
"grad_norm": 0.2253192514181137,
"learning_rate": 5e-05,
"loss": 1.8988468647003174,
"step": 1984
},
{
"epoch": 0.6925048381191485,
"grad_norm": 0.22085651755332947,
"learning_rate": 5e-05,
"loss": 2.098668336868286,
"step": 1986
},
{
"epoch": 0.6932022246630751,
"grad_norm": 0.2529553472995758,
"learning_rate": 5e-05,
"loss": 1.877577304840088,
"step": 1988
},
{
"epoch": 0.6938996112070017,
"grad_norm": 0.20500746369361877,
"learning_rate": 5e-05,
"loss": 1.9987249374389648,
"step": 1990
},
{
"epoch": 0.6945969977509284,
"grad_norm": 0.4637402594089508,
"learning_rate": 5e-05,
"loss": 1.9278911352157593,
"step": 1992
},
{
"epoch": 0.695294384294855,
"grad_norm": 0.243018239736557,
"learning_rate": 5e-05,
"loss": 1.951491117477417,
"step": 1994
},
{
"epoch": 0.6959917708387817,
"grad_norm": 0.22801651060581207,
"learning_rate": 5e-05,
"loss": 1.8762612342834473,
"step": 1996
},
{
"epoch": 0.6966891573827083,
"grad_norm": 0.2254435420036316,
"learning_rate": 5e-05,
"loss": 1.997175931930542,
"step": 1998
},
{
"epoch": 0.6973865439266349,
"grad_norm": 0.17995749413967133,
"learning_rate": 5e-05,
"loss": 1.8036714792251587,
"step": 2000
},
{
"epoch": 0.6980839304705616,
"grad_norm": 0.3124207556247711,
"learning_rate": 5e-05,
"loss": 1.9937630891799927,
"step": 2002
},
{
"epoch": 0.6987813170144882,
"grad_norm": 0.20657879114151,
"learning_rate": 5e-05,
"loss": 2.1100950241088867,
"step": 2004
},
{
"epoch": 0.6994787035584148,
"grad_norm": 0.25729361176490784,
"learning_rate": 5e-05,
"loss": 1.8586134910583496,
"step": 2006
},
{
"epoch": 0.7001760901023415,
"grad_norm": 0.17236414551734924,
"learning_rate": 5e-05,
"loss": 2.2024800777435303,
"step": 2008
},
{
"epoch": 0.7008734766462681,
"grad_norm": 0.19207501411437988,
"learning_rate": 5e-05,
"loss": 1.9439852237701416,
"step": 2010
},
{
"epoch": 0.7015708631901948,
"grad_norm": 0.20738041400909424,
"learning_rate": 5e-05,
"loss": 1.8800491094589233,
"step": 2012
},
{
"epoch": 0.7022682497341214,
"grad_norm": 0.17762541770935059,
"learning_rate": 5e-05,
"loss": 1.9378938674926758,
"step": 2014
},
{
"epoch": 0.702965636278048,
"grad_norm": 0.2435765415430069,
"learning_rate": 5e-05,
"loss": 2.0488927364349365,
"step": 2016
},
{
"epoch": 0.7036630228219747,
"grad_norm": 0.25711968541145325,
"learning_rate": 5e-05,
"loss": 1.8337585926055908,
"step": 2018
},
{
"epoch": 0.7043604093659013,
"grad_norm": 0.272570937871933,
"learning_rate": 5e-05,
"loss": 1.7698662281036377,
"step": 2020
},
{
"epoch": 0.7050577959098279,
"grad_norm": 0.1944895088672638,
"learning_rate": 5e-05,
"loss": 1.9470655918121338,
"step": 2022
},
{
"epoch": 0.7057551824537546,
"grad_norm": 0.20459075272083282,
"learning_rate": 5e-05,
"loss": 1.9057765007019043,
"step": 2024
},
{
"epoch": 0.7064525689976812,
"grad_norm": 0.2992056906223297,
"learning_rate": 5e-05,
"loss": 1.9368441104888916,
"step": 2026
},
{
"epoch": 0.7071499555416079,
"grad_norm": 0.24994510412216187,
"learning_rate": 5e-05,
"loss": 1.867093563079834,
"step": 2028
},
{
"epoch": 0.7078473420855345,
"grad_norm": 0.18913321197032928,
"learning_rate": 5e-05,
"loss": 2.0641443729400635,
"step": 2030
},
{
"epoch": 0.708544728629461,
"grad_norm": 0.3258306384086609,
"learning_rate": 5e-05,
"loss": 1.8863120079040527,
"step": 2032
},
{
"epoch": 0.7092421151733878,
"grad_norm": 0.2451459914445877,
"learning_rate": 5e-05,
"loss": 1.8047946691513062,
"step": 2034
},
{
"epoch": 0.7099395017173143,
"grad_norm": 0.21692386269569397,
"learning_rate": 5e-05,
"loss": 1.965606451034546,
"step": 2036
},
{
"epoch": 0.710636888261241,
"grad_norm": 0.21653258800506592,
"learning_rate": 5e-05,
"loss": 2.1215691566467285,
"step": 2038
},
{
"epoch": 0.7113342748051676,
"grad_norm": 0.25049927830696106,
"learning_rate": 5e-05,
"loss": 1.8238928318023682,
"step": 2040
},
{
"epoch": 0.7120316613490942,
"grad_norm": 0.23914234340190887,
"learning_rate": 5e-05,
"loss": 1.9403655529022217,
"step": 2042
},
{
"epoch": 0.7127290478930209,
"grad_norm": 0.2210332304239273,
"learning_rate": 5e-05,
"loss": 1.9609662294387817,
"step": 2044
},
{
"epoch": 0.7134264344369475,
"grad_norm": 0.25324663519859314,
"learning_rate": 5e-05,
"loss": 2.0118846893310547,
"step": 2046
},
{
"epoch": 0.7141238209808741,
"grad_norm": 0.23050567507743835,
"learning_rate": 5e-05,
"loss": 2.155590295791626,
"step": 2048
},
{
"epoch": 0.7148212075248008,
"grad_norm": 0.27442628145217896,
"learning_rate": 5e-05,
"loss": 1.8701404333114624,
"step": 2050
},
{
"epoch": 0.7155185940687274,
"grad_norm": 0.19824662804603577,
"learning_rate": 5e-05,
"loss": 1.9632997512817383,
"step": 2052
},
{
"epoch": 0.7162159806126541,
"grad_norm": 0.17668820917606354,
"learning_rate": 5e-05,
"loss": 2.0120043754577637,
"step": 2054
},
{
"epoch": 0.7169133671565807,
"grad_norm": 0.231851264834404,
"learning_rate": 5e-05,
"loss": 2.0922467708587646,
"step": 2056
},
{
"epoch": 0.7176107537005073,
"grad_norm": 0.2536415159702301,
"learning_rate": 5e-05,
"loss": 1.9578828811645508,
"step": 2058
},
{
"epoch": 0.718308140244434,
"grad_norm": 0.20625193417072296,
"learning_rate": 5e-05,
"loss": 2.0224688053131104,
"step": 2060
},
{
"epoch": 0.7190055267883606,
"grad_norm": 0.16995298862457275,
"learning_rate": 5e-05,
"loss": 1.9539906978607178,
"step": 2062
},
{
"epoch": 0.7197029133322873,
"grad_norm": 0.21072250604629517,
"learning_rate": 5e-05,
"loss": 1.9282296895980835,
"step": 2064
},
{
"epoch": 0.7204002998762139,
"grad_norm": 0.16904211044311523,
"learning_rate": 5e-05,
"loss": 2.323582649230957,
"step": 2066
},
{
"epoch": 0.7210976864201405,
"grad_norm": 0.20240511000156403,
"learning_rate": 5e-05,
"loss": 1.7976229190826416,
"step": 2068
},
{
"epoch": 0.7217950729640672,
"grad_norm": 0.21417611837387085,
"learning_rate": 5e-05,
"loss": 1.943137526512146,
"step": 2070
},
{
"epoch": 0.7224924595079938,
"grad_norm": 0.23913216590881348,
"learning_rate": 5e-05,
"loss": 1.892772912979126,
"step": 2072
},
{
"epoch": 0.7231898460519204,
"grad_norm": 0.210185244679451,
"learning_rate": 5e-05,
"loss": 2.1655025482177734,
"step": 2074
},
{
"epoch": 0.7238872325958471,
"grad_norm": 0.29644832015037537,
"learning_rate": 5e-05,
"loss": 1.9400770664215088,
"step": 2076
},
{
"epoch": 0.7245846191397737,
"grad_norm": 0.21575401723384857,
"learning_rate": 5e-05,
"loss": 1.8323345184326172,
"step": 2078
},
{
"epoch": 0.7252820056837004,
"grad_norm": 0.26795539259910583,
"learning_rate": 5e-05,
"loss": 1.9135476350784302,
"step": 2080
},
{
"epoch": 0.725979392227627,
"grad_norm": 0.21470215916633606,
"learning_rate": 5e-05,
"loss": 1.8226373195648193,
"step": 2082
},
{
"epoch": 0.7266767787715536,
"grad_norm": 0.23065048456192017,
"learning_rate": 5e-05,
"loss": 1.7718195915222168,
"step": 2084
},
{
"epoch": 0.7273741653154803,
"grad_norm": 0.23382049798965454,
"learning_rate": 5e-05,
"loss": 1.9421563148498535,
"step": 2086
},
{
"epoch": 0.7280715518594069,
"grad_norm": 0.20786328613758087,
"learning_rate": 5e-05,
"loss": 1.8556504249572754,
"step": 2088
},
{
"epoch": 0.7287689384033335,
"grad_norm": 0.2664857506752014,
"learning_rate": 5e-05,
"loss": 2.020401954650879,
"step": 2090
},
{
"epoch": 0.7294663249472602,
"grad_norm": 0.18101157248020172,
"learning_rate": 5e-05,
"loss": 1.9485325813293457,
"step": 2092
},
{
"epoch": 0.7301637114911868,
"grad_norm": 0.19057531654834747,
"learning_rate": 5e-05,
"loss": 1.931412696838379,
"step": 2094
},
{
"epoch": 0.7308610980351135,
"grad_norm": 0.20870794355869293,
"learning_rate": 5e-05,
"loss": 1.9227871894836426,
"step": 2096
},
{
"epoch": 0.73155848457904,
"grad_norm": 0.2855563461780548,
"learning_rate": 5e-05,
"loss": 1.9840281009674072,
"step": 2098
},
{
"epoch": 0.7322558711229666,
"grad_norm": 0.2462751567363739,
"learning_rate": 5e-05,
"loss": 1.9993574619293213,
"step": 2100
},
{
"epoch": 0.7329532576668933,
"grad_norm": 0.2574477791786194,
"learning_rate": 5e-05,
"loss": 2.060620069503784,
"step": 2102
},
{
"epoch": 0.7336506442108199,
"grad_norm": 0.1800810545682907,
"learning_rate": 5e-05,
"loss": 2.0911858081817627,
"step": 2104
},
{
"epoch": 0.7343480307547466,
"grad_norm": 0.17857886850833893,
"learning_rate": 5e-05,
"loss": 1.7908296585083008,
"step": 2106
},
{
"epoch": 0.7350454172986732,
"grad_norm": 0.2516303062438965,
"learning_rate": 5e-05,
"loss": 1.7887701988220215,
"step": 2108
},
{
"epoch": 0.7357428038425998,
"grad_norm": 0.20800773799419403,
"learning_rate": 5e-05,
"loss": 1.734657883644104,
"step": 2110
},
{
"epoch": 0.7364401903865265,
"grad_norm": 0.19409075379371643,
"learning_rate": 5e-05,
"loss": 1.9594202041625977,
"step": 2112
},
{
"epoch": 0.7371375769304531,
"grad_norm": 0.17825016379356384,
"learning_rate": 5e-05,
"loss": 1.9487360715866089,
"step": 2114
},
{
"epoch": 0.7378349634743797,
"grad_norm": 0.18024425208568573,
"learning_rate": 5e-05,
"loss": 2.1237568855285645,
"step": 2116
},
{
"epoch": 0.7385323500183064,
"grad_norm": 0.22989214956760406,
"learning_rate": 5e-05,
"loss": 1.931516408920288,
"step": 2118
},
{
"epoch": 0.739229736562233,
"grad_norm": 0.20000092685222626,
"learning_rate": 5e-05,
"loss": 2.03963041305542,
"step": 2120
},
{
"epoch": 0.7399271231061597,
"grad_norm": 0.24111893773078918,
"learning_rate": 5e-05,
"loss": 2.027360439300537,
"step": 2122
},
{
"epoch": 0.7406245096500863,
"grad_norm": 0.20106832683086395,
"learning_rate": 5e-05,
"loss": 1.915521502494812,
"step": 2124
},
{
"epoch": 0.7413218961940129,
"grad_norm": 0.2090524286031723,
"learning_rate": 5e-05,
"loss": 1.9295654296875,
"step": 2126
},
{
"epoch": 0.7420192827379396,
"grad_norm": 0.176246777176857,
"learning_rate": 5e-05,
"loss": 1.9641776084899902,
"step": 2128
},
{
"epoch": 0.7427166692818662,
"grad_norm": 0.24356651306152344,
"learning_rate": 5e-05,
"loss": 1.9602073431015015,
"step": 2130
},
{
"epoch": 0.7434140558257929,
"grad_norm": 0.18876545131206512,
"learning_rate": 5e-05,
"loss": 2.029855489730835,
"step": 2132
},
{
"epoch": 0.7441114423697195,
"grad_norm": 0.21620714664459229,
"learning_rate": 5e-05,
"loss": 2.163750648498535,
"step": 2134
},
{
"epoch": 0.7448088289136461,
"grad_norm": 0.21363040804862976,
"learning_rate": 5e-05,
"loss": 1.8749510049819946,
"step": 2136
},
{
"epoch": 0.7455062154575728,
"grad_norm": 0.2743799090385437,
"learning_rate": 5e-05,
"loss": 2.015801191329956,
"step": 2138
},
{
"epoch": 0.7462036020014994,
"grad_norm": 0.259393572807312,
"learning_rate": 5e-05,
"loss": 1.896141767501831,
"step": 2140
},
{
"epoch": 0.746900988545426,
"grad_norm": 0.19506627321243286,
"learning_rate": 5e-05,
"loss": 1.6852531433105469,
"step": 2142
},
{
"epoch": 0.7475983750893527,
"grad_norm": 0.2716883420944214,
"learning_rate": 5e-05,
"loss": 1.9007842540740967,
"step": 2144
},
{
"epoch": 0.7482957616332793,
"grad_norm": 0.23826134204864502,
"learning_rate": 5e-05,
"loss": 2.0209457874298096,
"step": 2146
},
{
"epoch": 0.748993148177206,
"grad_norm": 0.16911599040031433,
"learning_rate": 5e-05,
"loss": 1.9810214042663574,
"step": 2148
},
{
"epoch": 0.7496905347211326,
"grad_norm": 0.21925072371959686,
"learning_rate": 5e-05,
"loss": 2.1360256671905518,
"step": 2150
},
{
"epoch": 0.7503879212650592,
"grad_norm": 0.1755322813987732,
"learning_rate": 5e-05,
"loss": 2.0105040073394775,
"step": 2152
},
{
"epoch": 0.7510853078089859,
"grad_norm": 0.2572018504142761,
"learning_rate": 5e-05,
"loss": 1.8643507957458496,
"step": 2154
},
{
"epoch": 0.7517826943529125,
"grad_norm": 0.23412656784057617,
"learning_rate": 5e-05,
"loss": 2.103471040725708,
"step": 2156
},
{
"epoch": 0.752480080896839,
"grad_norm": 0.22634395956993103,
"learning_rate": 5e-05,
"loss": 2.0817770957946777,
"step": 2158
},
{
"epoch": 0.7531774674407657,
"grad_norm": 0.18061764538288116,
"learning_rate": 5e-05,
"loss": 1.993519902229309,
"step": 2160
},
{
"epoch": 0.7538748539846923,
"grad_norm": 0.2553674280643463,
"learning_rate": 5e-05,
"loss": 2.0109446048736572,
"step": 2162
},
{
"epoch": 0.754572240528619,
"grad_norm": 0.19744637608528137,
"learning_rate": 5e-05,
"loss": 1.9850050210952759,
"step": 2164
},
{
"epoch": 0.7552696270725456,
"grad_norm": 0.17519602179527283,
"learning_rate": 5e-05,
"loss": 2.0965726375579834,
"step": 2166
},
{
"epoch": 0.7559670136164722,
"grad_norm": 0.16751813888549805,
"learning_rate": 5e-05,
"loss": 1.9219980239868164,
"step": 2168
},
{
"epoch": 0.7566644001603989,
"grad_norm": 0.24096159636974335,
"learning_rate": 5e-05,
"loss": 1.8713510036468506,
"step": 2170
},
{
"epoch": 0.7573617867043255,
"grad_norm": 0.24218828976154327,
"learning_rate": 5e-05,
"loss": 1.9764506816864014,
"step": 2172
},
{
"epoch": 0.7580591732482522,
"grad_norm": 0.1704159528017044,
"learning_rate": 5e-05,
"loss": 1.879757285118103,
"step": 2174
},
{
"epoch": 0.7587565597921788,
"grad_norm": 0.19566938281059265,
"learning_rate": 5e-05,
"loss": 1.9094607830047607,
"step": 2176
},
{
"epoch": 0.7594539463361054,
"grad_norm": 0.21699845790863037,
"learning_rate": 5e-05,
"loss": 1.8993353843688965,
"step": 2178
},
{
"epoch": 0.7601513328800321,
"grad_norm": 0.25892215967178345,
"learning_rate": 5e-05,
"loss": 1.811294436454773,
"step": 2180
},
{
"epoch": 0.7608487194239587,
"grad_norm": 0.23539891839027405,
"learning_rate": 5e-05,
"loss": 1.9238123893737793,
"step": 2182
},
{
"epoch": 0.7615461059678853,
"grad_norm": 0.20322979986667633,
"learning_rate": 5e-05,
"loss": 1.942223072052002,
"step": 2184
},
{
"epoch": 0.762243492511812,
"grad_norm": 0.1692976951599121,
"learning_rate": 5e-05,
"loss": 1.976067066192627,
"step": 2186
},
{
"epoch": 0.7629408790557386,
"grad_norm": 0.2596499025821686,
"learning_rate": 5e-05,
"loss": 1.8981826305389404,
"step": 2188
},
{
"epoch": 0.7636382655996653,
"grad_norm": 0.17991489171981812,
"learning_rate": 5e-05,
"loss": 2.1444361209869385,
"step": 2190
},
{
"epoch": 0.7643356521435919,
"grad_norm": 0.19914187490940094,
"learning_rate": 5e-05,
"loss": 2.042095422744751,
"step": 2192
},
{
"epoch": 0.7650330386875185,
"grad_norm": 0.2698531448841095,
"learning_rate": 5e-05,
"loss": 1.787888765335083,
"step": 2194
},
{
"epoch": 0.7657304252314452,
"grad_norm": 0.2817508578300476,
"learning_rate": 5e-05,
"loss": 2.0970849990844727,
"step": 2196
},
{
"epoch": 0.7664278117753718,
"grad_norm": 0.23461763560771942,
"learning_rate": 5e-05,
"loss": 1.9004498720169067,
"step": 2198
},
{
"epoch": 0.7671251983192984,
"grad_norm": 0.2059352993965149,
"learning_rate": 5e-05,
"loss": 1.7717727422714233,
"step": 2200
},
{
"epoch": 0.7678225848632251,
"grad_norm": 0.2654406726360321,
"learning_rate": 5e-05,
"loss": 1.8539447784423828,
"step": 2202
},
{
"epoch": 0.7685199714071517,
"grad_norm": 0.18033595383167267,
"learning_rate": 5e-05,
"loss": 1.951402187347412,
"step": 2204
},
{
"epoch": 0.7692173579510784,
"grad_norm": 0.2133253514766693,
"learning_rate": 5e-05,
"loss": 1.9016897678375244,
"step": 2206
},
{
"epoch": 0.769914744495005,
"grad_norm": 0.33630532026290894,
"learning_rate": 5e-05,
"loss": 1.8270273208618164,
"step": 2208
},
{
"epoch": 0.7706121310389316,
"grad_norm": 0.2575637698173523,
"learning_rate": 5e-05,
"loss": 1.9425978660583496,
"step": 2210
},
{
"epoch": 0.7713095175828583,
"grad_norm": 0.20577408373355865,
"learning_rate": 5e-05,
"loss": 1.9617164134979248,
"step": 2212
},
{
"epoch": 0.7720069041267849,
"grad_norm": 0.20532967150211334,
"learning_rate": 5e-05,
"loss": 2.038666248321533,
"step": 2214
},
{
"epoch": 0.7727042906707116,
"grad_norm": 0.18110215663909912,
"learning_rate": 5e-05,
"loss": 1.6460938453674316,
"step": 2216
},
{
"epoch": 0.7734016772146382,
"grad_norm": 0.21534425020217896,
"learning_rate": 5e-05,
"loss": 1.9891645908355713,
"step": 2218
},
{
"epoch": 0.7740990637585647,
"grad_norm": 0.16367346048355103,
"learning_rate": 5e-05,
"loss": 2.038942813873291,
"step": 2220
},
{
"epoch": 0.7747964503024914,
"grad_norm": 0.2003421038389206,
"learning_rate": 5e-05,
"loss": 1.862746238708496,
"step": 2222
},
{
"epoch": 0.775493836846418,
"grad_norm": 0.20611684024333954,
"learning_rate": 5e-05,
"loss": 1.5451436042785645,
"step": 2224
},
{
"epoch": 0.7761912233903446,
"grad_norm": 0.24627675116062164,
"learning_rate": 5e-05,
"loss": 1.7264751195907593,
"step": 2226
},
{
"epoch": 0.7768886099342713,
"grad_norm": 0.26181432604789734,
"learning_rate": 5e-05,
"loss": 1.7415522336959839,
"step": 2228
},
{
"epoch": 0.7775859964781979,
"grad_norm": 0.1756943166255951,
"learning_rate": 5e-05,
"loss": 1.9747660160064697,
"step": 2230
},
{
"epoch": 0.7782833830221246,
"grad_norm": 0.21931739151477814,
"learning_rate": 5e-05,
"loss": 1.9190582036972046,
"step": 2232
},
{
"epoch": 0.7789807695660512,
"grad_norm": 0.24526743590831757,
"learning_rate": 5e-05,
"loss": 1.6226704120635986,
"step": 2234
},
{
"epoch": 0.7796781561099778,
"grad_norm": 0.21926890313625336,
"learning_rate": 5e-05,
"loss": 1.9680263996124268,
"step": 2236
},
{
"epoch": 0.7803755426539045,
"grad_norm": 0.3071652948856354,
"learning_rate": 5e-05,
"loss": 1.639426350593567,
"step": 2238
},
{
"epoch": 0.7810729291978311,
"grad_norm": 0.20577116310596466,
"learning_rate": 5e-05,
"loss": 1.8001503944396973,
"step": 2240
},
{
"epoch": 0.7817703157417578,
"grad_norm": 0.1559043526649475,
"learning_rate": 5e-05,
"loss": 1.6160184144973755,
"step": 2242
},
{
"epoch": 0.7824677022856844,
"grad_norm": 0.2791428565979004,
"learning_rate": 5e-05,
"loss": 1.816514492034912,
"step": 2244
},
{
"epoch": 0.783165088829611,
"grad_norm": 0.21342822909355164,
"learning_rate": 5e-05,
"loss": 1.8380680084228516,
"step": 2246
},
{
"epoch": 0.7838624753735377,
"grad_norm": 0.19591249525547028,
"learning_rate": 5e-05,
"loss": 1.8789033889770508,
"step": 2248
},
{
"epoch": 0.7845598619174643,
"grad_norm": 0.23627856373786926,
"learning_rate": 5e-05,
"loss": 2.068758726119995,
"step": 2250
},
{
"epoch": 0.7852572484613909,
"grad_norm": 0.28036272525787354,
"learning_rate": 5e-05,
"loss": 1.8976188898086548,
"step": 2252
},
{
"epoch": 0.7859546350053176,
"grad_norm": 0.24042636156082153,
"learning_rate": 5e-05,
"loss": 1.973801612854004,
"step": 2254
},
{
"epoch": 0.7866520215492442,
"grad_norm": 0.19923457503318787,
"learning_rate": 5e-05,
"loss": 1.9272568225860596,
"step": 2256
},
{
"epoch": 0.7873494080931709,
"grad_norm": 0.1986333131790161,
"learning_rate": 5e-05,
"loss": 1.8647773265838623,
"step": 2258
},
{
"epoch": 0.7880467946370975,
"grad_norm": 0.2959451675415039,
"learning_rate": 5e-05,
"loss": 1.822899341583252,
"step": 2260
},
{
"epoch": 0.7887441811810241,
"grad_norm": 0.21408234536647797,
"learning_rate": 5e-05,
"loss": 2.083057403564453,
"step": 2262
},
{
"epoch": 0.7894415677249508,
"grad_norm": 0.24847257137298584,
"learning_rate": 5e-05,
"loss": 2.041271686553955,
"step": 2264
},
{
"epoch": 0.7901389542688774,
"grad_norm": 0.2285429835319519,
"learning_rate": 5e-05,
"loss": 1.8865199089050293,
"step": 2266
},
{
"epoch": 0.790836340812804,
"grad_norm": 0.2242038995027542,
"learning_rate": 5e-05,
"loss": 1.7470977306365967,
"step": 2268
},
{
"epoch": 0.7915337273567307,
"grad_norm": 0.23584255576133728,
"learning_rate": 5e-05,
"loss": 1.9168062210083008,
"step": 2270
},
{
"epoch": 0.7922311139006573,
"grad_norm": 0.18825343251228333,
"learning_rate": 5e-05,
"loss": 1.99519944190979,
"step": 2272
},
{
"epoch": 0.792928500444584,
"grad_norm": 0.31738072633743286,
"learning_rate": 5e-05,
"loss": 1.750924825668335,
"step": 2274
},
{
"epoch": 0.7936258869885106,
"grad_norm": 0.1915413737297058,
"learning_rate": 5e-05,
"loss": 2.1374001502990723,
"step": 2276
},
{
"epoch": 0.7943232735324371,
"grad_norm": 0.20662441849708557,
"learning_rate": 5e-05,
"loss": 1.8818833827972412,
"step": 2278
},
{
"epoch": 0.7950206600763638,
"grad_norm": 0.2140003889799118,
"learning_rate": 5e-05,
"loss": 2.065359115600586,
"step": 2280
},
{
"epoch": 0.7957180466202904,
"grad_norm": 0.22776512801647186,
"learning_rate": 5e-05,
"loss": 1.9953523874282837,
"step": 2282
},
{
"epoch": 0.7964154331642171,
"grad_norm": 0.3272683918476105,
"learning_rate": 5e-05,
"loss": 1.8373709917068481,
"step": 2284
},
{
"epoch": 0.7971128197081437,
"grad_norm": 0.278938889503479,
"learning_rate": 5e-05,
"loss": 1.9822043180465698,
"step": 2286
},
{
"epoch": 0.7978102062520703,
"grad_norm": 0.2507492005825043,
"learning_rate": 5e-05,
"loss": 2.0197887420654297,
"step": 2288
},
{
"epoch": 0.798507592795997,
"grad_norm": 0.29306724667549133,
"learning_rate": 5e-05,
"loss": 1.9630277156829834,
"step": 2290
},
{
"epoch": 0.7992049793399236,
"grad_norm": 0.22105816006660461,
"learning_rate": 5e-05,
"loss": 2.0450775623321533,
"step": 2292
},
{
"epoch": 0.7999023658838502,
"grad_norm": 0.3041655421257019,
"learning_rate": 5e-05,
"loss": 1.637634038925171,
"step": 2294
},
{
"epoch": 0.8005997524277769,
"grad_norm": 0.2233806550502777,
"learning_rate": 5e-05,
"loss": 1.8702466487884521,
"step": 2296
},
{
"epoch": 0.8012971389717035,
"grad_norm": 0.34665629267692566,
"learning_rate": 5e-05,
"loss": 1.5512590408325195,
"step": 2298
},
{
"epoch": 0.8019945255156302,
"grad_norm": 0.2618155777454376,
"learning_rate": 5e-05,
"loss": 1.8056235313415527,
"step": 2300
},
{
"epoch": 0.8026919120595568,
"grad_norm": 0.20531056821346283,
"learning_rate": 5e-05,
"loss": 1.9945034980773926,
"step": 2302
},
{
"epoch": 0.8033892986034834,
"grad_norm": 0.25889530777931213,
"learning_rate": 5e-05,
"loss": 1.8921455144882202,
"step": 2304
},
{
"epoch": 0.8040866851474101,
"grad_norm": 0.1906490921974182,
"learning_rate": 5e-05,
"loss": 1.982350468635559,
"step": 2306
},
{
"epoch": 0.8047840716913367,
"grad_norm": 0.3171706795692444,
"learning_rate": 5e-05,
"loss": 1.8985872268676758,
"step": 2308
},
{
"epoch": 0.8054814582352634,
"grad_norm": 0.18568949401378632,
"learning_rate": 5e-05,
"loss": 1.9411725997924805,
"step": 2310
},
{
"epoch": 0.80617884477919,
"grad_norm": 0.2307809740304947,
"learning_rate": 5e-05,
"loss": 1.8542625904083252,
"step": 2312
},
{
"epoch": 0.8068762313231166,
"grad_norm": 0.29474905133247375,
"learning_rate": 5e-05,
"loss": 1.7415797710418701,
"step": 2314
},
{
"epoch": 0.8075736178670433,
"grad_norm": 0.1813082993030548,
"learning_rate": 5e-05,
"loss": 1.8530118465423584,
"step": 2316
},
{
"epoch": 0.8082710044109699,
"grad_norm": 0.30233779549598694,
"learning_rate": 5e-05,
"loss": 1.732279658317566,
"step": 2318
},
{
"epoch": 0.8089683909548965,
"grad_norm": 0.19910310208797455,
"learning_rate": 5e-05,
"loss": 1.7143280506134033,
"step": 2320
},
{
"epoch": 0.8096657774988232,
"grad_norm": 0.2825470268726349,
"learning_rate": 5e-05,
"loss": 1.8623201847076416,
"step": 2322
},
{
"epoch": 0.8103631640427498,
"grad_norm": 0.1972043663263321,
"learning_rate": 5e-05,
"loss": 2.0700478553771973,
"step": 2324
},
{
"epoch": 0.8110605505866765,
"grad_norm": 0.29088762402534485,
"learning_rate": 5e-05,
"loss": 1.7995343208312988,
"step": 2326
},
{
"epoch": 0.8117579371306031,
"grad_norm": 0.2989962697029114,
"learning_rate": 5e-05,
"loss": 1.6743077039718628,
"step": 2328
},
{
"epoch": 0.8124553236745297,
"grad_norm": 0.24003519117832184,
"learning_rate": 5e-05,
"loss": 1.9525628089904785,
"step": 2330
},
{
"epoch": 0.8131527102184564,
"grad_norm": 0.21295663714408875,
"learning_rate": 5e-05,
"loss": 1.7714589834213257,
"step": 2332
},
{
"epoch": 0.813850096762383,
"grad_norm": 0.23063835501670837,
"learning_rate": 5e-05,
"loss": 1.7743947505950928,
"step": 2334
},
{
"epoch": 0.8145474833063095,
"grad_norm": 0.21624773740768433,
"learning_rate": 5e-05,
"loss": 1.870970368385315,
"step": 2336
},
{
"epoch": 0.8152448698502363,
"grad_norm": 0.34053564071655273,
"learning_rate": 5e-05,
"loss": 1.5913314819335938,
"step": 2338
},
{
"epoch": 0.8159422563941628,
"grad_norm": 0.20439667999744415,
"learning_rate": 5e-05,
"loss": 1.9366562366485596,
"step": 2340
},
{
"epoch": 0.8166396429380895,
"grad_norm": 0.22627848386764526,
"learning_rate": 5e-05,
"loss": 1.841552495956421,
"step": 2342
},
{
"epoch": 0.8173370294820161,
"grad_norm": 0.1971031129360199,
"learning_rate": 5e-05,
"loss": 1.6385873556137085,
"step": 2344
},
{
"epoch": 0.8180344160259427,
"grad_norm": 0.2172694057226181,
"learning_rate": 5e-05,
"loss": 1.9489915370941162,
"step": 2346
},
{
"epoch": 0.8187318025698694,
"grad_norm": 0.4004741609096527,
"learning_rate": 5e-05,
"loss": 1.983567714691162,
"step": 2348
},
{
"epoch": 0.819429189113796,
"grad_norm": 0.18114428222179413,
"learning_rate": 5e-05,
"loss": 1.7686386108398438,
"step": 2350
},
{
"epoch": 0.8201265756577227,
"grad_norm": 0.1820041537284851,
"learning_rate": 5e-05,
"loss": 1.9830634593963623,
"step": 2352
},
{
"epoch": 0.8208239622016493,
"grad_norm": 0.20078644156455994,
"learning_rate": 5e-05,
"loss": 1.6137222051620483,
"step": 2354
},
{
"epoch": 0.8215213487455759,
"grad_norm": 0.20240375399589539,
"learning_rate": 5e-05,
"loss": 1.7961931228637695,
"step": 2356
},
{
"epoch": 0.8222187352895026,
"grad_norm": 0.2730301022529602,
"learning_rate": 5e-05,
"loss": 1.9427998065948486,
"step": 2358
},
{
"epoch": 0.8229161218334292,
"grad_norm": 0.20543934404850006,
"learning_rate": 5e-05,
"loss": 1.8742575645446777,
"step": 2360
},
{
"epoch": 0.8236135083773558,
"grad_norm": 0.2649659514427185,
"learning_rate": 5e-05,
"loss": 1.9555909633636475,
"step": 2362
},
{
"epoch": 0.8243108949212825,
"grad_norm": 0.2065000981092453,
"learning_rate": 5e-05,
"loss": 1.8060014247894287,
"step": 2364
},
{
"epoch": 0.8250082814652091,
"grad_norm": 0.2397371530532837,
"learning_rate": 5e-05,
"loss": 2.059617042541504,
"step": 2366
},
{
"epoch": 0.8257056680091358,
"grad_norm": 0.20603303611278534,
"learning_rate": 5e-05,
"loss": 1.8409523963928223,
"step": 2368
},
{
"epoch": 0.8264030545530624,
"grad_norm": 0.2540438175201416,
"learning_rate": 5e-05,
"loss": 1.8821148872375488,
"step": 2370
},
{
"epoch": 0.827100441096989,
"grad_norm": 0.2622252404689789,
"learning_rate": 5e-05,
"loss": 1.8382006883621216,
"step": 2372
},
{
"epoch": 0.8277978276409157,
"grad_norm": 0.26272404193878174,
"learning_rate": 5e-05,
"loss": 1.9710807800292969,
"step": 2374
},
{
"epoch": 0.8284952141848423,
"grad_norm": 0.19015145301818848,
"learning_rate": 5e-05,
"loss": 1.869012713432312,
"step": 2376
},
{
"epoch": 0.829192600728769,
"grad_norm": 0.2541094422340393,
"learning_rate": 5e-05,
"loss": 1.9601718187332153,
"step": 2378
},
{
"epoch": 0.8298899872726956,
"grad_norm": 0.265034019947052,
"learning_rate": 5e-05,
"loss": 1.7983336448669434,
"step": 2380
},
{
"epoch": 0.8305873738166222,
"grad_norm": 0.28119009733200073,
"learning_rate": 5e-05,
"loss": 1.9807533025741577,
"step": 2382
},
{
"epoch": 0.8312847603605489,
"grad_norm": 0.18375326693058014,
"learning_rate": 5e-05,
"loss": 1.903433084487915,
"step": 2384
},
{
"epoch": 0.8319821469044755,
"grad_norm": 0.26467451453208923,
"learning_rate": 5e-05,
"loss": 1.8559367656707764,
"step": 2386
},
{
"epoch": 0.8326795334484021,
"grad_norm": 0.21322381496429443,
"learning_rate": 5e-05,
"loss": 2.0329360961914062,
"step": 2388
},
{
"epoch": 0.8333769199923288,
"grad_norm": 0.19978481531143188,
"learning_rate": 5e-05,
"loss": 1.928691029548645,
"step": 2390
},
{
"epoch": 0.8340743065362554,
"grad_norm": 0.26044854521751404,
"learning_rate": 5e-05,
"loss": 2.0241265296936035,
"step": 2392
},
{
"epoch": 0.8347716930801821,
"grad_norm": 0.25593796372413635,
"learning_rate": 5e-05,
"loss": 1.8040621280670166,
"step": 2394
},
{
"epoch": 0.8354690796241087,
"grad_norm": 0.2556050419807434,
"learning_rate": 5e-05,
"loss": 1.8471500873565674,
"step": 2396
},
{
"epoch": 0.8361664661680352,
"grad_norm": 0.29592692852020264,
"learning_rate": 5e-05,
"loss": 1.8190295696258545,
"step": 2398
},
{
"epoch": 0.836863852711962,
"grad_norm": 0.22786323726177216,
"learning_rate": 5e-05,
"loss": 1.8059440851211548,
"step": 2400
},
{
"epoch": 0.8375612392558885,
"grad_norm": 0.21543928980827332,
"learning_rate": 5e-05,
"loss": 1.8659863471984863,
"step": 2402
},
{
"epoch": 0.8382586257998151,
"grad_norm": 0.24894928932189941,
"learning_rate": 5e-05,
"loss": 2.1881353855133057,
"step": 2404
},
{
"epoch": 0.8389560123437418,
"grad_norm": 0.2092692255973816,
"learning_rate": 5e-05,
"loss": 1.8489896059036255,
"step": 2406
},
{
"epoch": 0.8396533988876684,
"grad_norm": 0.297468900680542,
"learning_rate": 5e-05,
"loss": 2.0073957443237305,
"step": 2408
},
{
"epoch": 0.8403507854315951,
"grad_norm": 0.19659481942653656,
"learning_rate": 5e-05,
"loss": 1.917393684387207,
"step": 2410
},
{
"epoch": 0.8410481719755217,
"grad_norm": 0.3369445204734802,
"learning_rate": 5e-05,
"loss": 1.8730225563049316,
"step": 2412
},
{
"epoch": 0.8417455585194483,
"grad_norm": 0.14724263548851013,
"learning_rate": 5e-05,
"loss": 2.0430855751037598,
"step": 2414
},
{
"epoch": 0.842442945063375,
"grad_norm": 0.19056470692157745,
"learning_rate": 5e-05,
"loss": 1.980102300643921,
"step": 2416
},
{
"epoch": 0.8431403316073016,
"grad_norm": 0.1984318494796753,
"learning_rate": 5e-05,
"loss": 2.017703056335449,
"step": 2418
},
{
"epoch": 0.8438377181512283,
"grad_norm": 0.1699439436197281,
"learning_rate": 5e-05,
"loss": 1.795714020729065,
"step": 2420
},
{
"epoch": 0.8445351046951549,
"grad_norm": 0.29702281951904297,
"learning_rate": 5e-05,
"loss": 2.1077191829681396,
"step": 2422
},
{
"epoch": 0.8452324912390815,
"grad_norm": 0.22882135212421417,
"learning_rate": 5e-05,
"loss": 1.92616605758667,
"step": 2424
},
{
"epoch": 0.8459298777830082,
"grad_norm": 0.21170425415039062,
"learning_rate": 5e-05,
"loss": 1.9746460914611816,
"step": 2426
},
{
"epoch": 0.8466272643269348,
"grad_norm": 0.21631506085395813,
"learning_rate": 5e-05,
"loss": 1.7190144062042236,
"step": 2428
},
{
"epoch": 0.8473246508708614,
"grad_norm": 0.21133138239383698,
"learning_rate": 5e-05,
"loss": 1.7343809604644775,
"step": 2430
},
{
"epoch": 0.8480220374147881,
"grad_norm": 0.20639613270759583,
"learning_rate": 5e-05,
"loss": 1.9387023448944092,
"step": 2432
},
{
"epoch": 0.8487194239587147,
"grad_norm": 0.24656221270561218,
"learning_rate": 5e-05,
"loss": 1.852571725845337,
"step": 2434
},
{
"epoch": 0.8494168105026414,
"grad_norm": 0.18058742582798004,
"learning_rate": 5e-05,
"loss": 2.081359624862671,
"step": 2436
},
{
"epoch": 0.850114197046568,
"grad_norm": 0.18465012311935425,
"learning_rate": 5e-05,
"loss": 1.93831467628479,
"step": 2438
},
{
"epoch": 0.8508115835904946,
"grad_norm": 0.2825033962726593,
"learning_rate": 5e-05,
"loss": 1.9698466062545776,
"step": 2440
},
{
"epoch": 0.8515089701344213,
"grad_norm": 0.27780500054359436,
"learning_rate": 5e-05,
"loss": 1.982361078262329,
"step": 2442
},
{
"epoch": 0.8522063566783479,
"grad_norm": 0.16493284702301025,
"learning_rate": 5e-05,
"loss": 1.9893501996994019,
"step": 2444
},
{
"epoch": 0.8529037432222746,
"grad_norm": 0.1815342903137207,
"learning_rate": 5e-05,
"loss": 1.9853184223175049,
"step": 2446
},
{
"epoch": 0.8536011297662012,
"grad_norm": 0.19716495275497437,
"learning_rate": 5e-05,
"loss": 1.7325427532196045,
"step": 2448
},
{
"epoch": 0.8542985163101278,
"grad_norm": 0.2385205179452896,
"learning_rate": 5e-05,
"loss": 1.9826991558074951,
"step": 2450
},
{
"epoch": 0.8549959028540545,
"grad_norm": 0.19368870556354523,
"learning_rate": 5e-05,
"loss": 1.8701865673065186,
"step": 2452
},
{
"epoch": 0.8556932893979811,
"grad_norm": 0.17733433842658997,
"learning_rate": 5e-05,
"loss": 2.036306381225586,
"step": 2454
},
{
"epoch": 0.8563906759419077,
"grad_norm": 0.2510242164134979,
"learning_rate": 5e-05,
"loss": 1.9895774126052856,
"step": 2456
},
{
"epoch": 0.8570880624858344,
"grad_norm": 0.21947576105594635,
"learning_rate": 5e-05,
"loss": 1.9222482442855835,
"step": 2458
},
{
"epoch": 0.857785449029761,
"grad_norm": 0.2402341365814209,
"learning_rate": 5e-05,
"loss": 2.0680880546569824,
"step": 2460
},
{
"epoch": 0.8584828355736877,
"grad_norm": 0.2076481431722641,
"learning_rate": 5e-05,
"loss": 1.7876285314559937,
"step": 2462
},
{
"epoch": 0.8591802221176142,
"grad_norm": 0.22248335182666779,
"learning_rate": 5e-05,
"loss": 1.746431827545166,
"step": 2464
},
{
"epoch": 0.8598776086615408,
"grad_norm": 0.2763570249080658,
"learning_rate": 5e-05,
"loss": 1.830052137374878,
"step": 2466
},
{
"epoch": 0.8605749952054675,
"grad_norm": 0.242806538939476,
"learning_rate": 5e-05,
"loss": 1.820050597190857,
"step": 2468
},
{
"epoch": 0.8612723817493941,
"grad_norm": 0.19415214657783508,
"learning_rate": 5e-05,
"loss": 1.8796792030334473,
"step": 2470
},
{
"epoch": 0.8619697682933207,
"grad_norm": 0.16487431526184082,
"learning_rate": 5e-05,
"loss": 1.8481247425079346,
"step": 2472
},
{
"epoch": 0.8626671548372474,
"grad_norm": 0.18796490132808685,
"learning_rate": 5e-05,
"loss": 2.0668163299560547,
"step": 2474
},
{
"epoch": 0.863364541381174,
"grad_norm": 0.16939391195774078,
"learning_rate": 5e-05,
"loss": 1.9822109937667847,
"step": 2476
},
{
"epoch": 0.8640619279251007,
"grad_norm": 0.22088705003261566,
"learning_rate": 5e-05,
"loss": 1.8946961164474487,
"step": 2478
},
{
"epoch": 0.8647593144690273,
"grad_norm": 0.19917675852775574,
"learning_rate": 5e-05,
"loss": 1.9240469932556152,
"step": 2480
},
{
"epoch": 0.8654567010129539,
"grad_norm": 0.1673891395330429,
"learning_rate": 5e-05,
"loss": 2.0632004737854004,
"step": 2482
},
{
"epoch": 0.8661540875568806,
"grad_norm": 0.18360723555088043,
"learning_rate": 5e-05,
"loss": 1.8349483013153076,
"step": 2484
},
{
"epoch": 0.8668514741008072,
"grad_norm": 0.2690318822860718,
"learning_rate": 5e-05,
"loss": 1.7954403162002563,
"step": 2486
},
{
"epoch": 0.8675488606447339,
"grad_norm": 0.19765828549861908,
"learning_rate": 5e-05,
"loss": 2.0164332389831543,
"step": 2488
},
{
"epoch": 0.8682462471886605,
"grad_norm": 0.26011911034584045,
"learning_rate": 5e-05,
"loss": 1.8244246244430542,
"step": 2490
},
{
"epoch": 0.8689436337325871,
"grad_norm": 0.22959889471530914,
"learning_rate": 5e-05,
"loss": 1.931602120399475,
"step": 2492
},
{
"epoch": 0.8696410202765138,
"grad_norm": 0.19653260707855225,
"learning_rate": 5e-05,
"loss": 1.7620649337768555,
"step": 2494
},
{
"epoch": 0.8703384068204404,
"grad_norm": 0.2762472927570343,
"learning_rate": 5e-05,
"loss": 1.8492257595062256,
"step": 2496
},
{
"epoch": 0.871035793364367,
"grad_norm": 0.22183212637901306,
"learning_rate": 5e-05,
"loss": 1.9420517683029175,
"step": 2498
},
{
"epoch": 0.8717331799082937,
"grad_norm": 0.24676363170146942,
"learning_rate": 5e-05,
"loss": 1.8644121885299683,
"step": 2500
},
{
"epoch": 0.8724305664522203,
"grad_norm": 0.2391708493232727,
"learning_rate": 5e-05,
"loss": 1.934468150138855,
"step": 2502
},
{
"epoch": 0.873127952996147,
"grad_norm": 0.15963231027126312,
"learning_rate": 5e-05,
"loss": 1.9977290630340576,
"step": 2504
},
{
"epoch": 0.8738253395400736,
"grad_norm": 0.23561052978038788,
"learning_rate": 5e-05,
"loss": 2.0387420654296875,
"step": 2506
},
{
"epoch": 0.8745227260840002,
"grad_norm": 0.3309169113636017,
"learning_rate": 5e-05,
"loss": 1.9751169681549072,
"step": 2508
},
{
"epoch": 0.8752201126279269,
"grad_norm": 0.26138627529144287,
"learning_rate": 5e-05,
"loss": 1.903587818145752,
"step": 2510
},
{
"epoch": 0.8759174991718535,
"grad_norm": 0.20577403903007507,
"learning_rate": 5e-05,
"loss": 1.7635977268218994,
"step": 2512
},
{
"epoch": 0.8766148857157802,
"grad_norm": 0.2254587560892105,
"learning_rate": 5e-05,
"loss": 1.7151050567626953,
"step": 2514
},
{
"epoch": 0.8773122722597068,
"grad_norm": 0.16640746593475342,
"learning_rate": 5e-05,
"loss": 1.6992413997650146,
"step": 2516
},
{
"epoch": 0.8780096588036334,
"grad_norm": 0.252765029668808,
"learning_rate": 5e-05,
"loss": 1.9397852420806885,
"step": 2518
},
{
"epoch": 0.8787070453475601,
"grad_norm": 0.21200324594974518,
"learning_rate": 5e-05,
"loss": 1.9401975870132446,
"step": 2520
},
{
"epoch": 0.8794044318914866,
"grad_norm": 0.15726792812347412,
"learning_rate": 5e-05,
"loss": 2.076117992401123,
"step": 2522
},
{
"epoch": 0.8801018184354132,
"grad_norm": 0.2524771988391876,
"learning_rate": 5e-05,
"loss": 1.652165174484253,
"step": 2524
},
{
"epoch": 0.88079920497934,
"grad_norm": 0.23599308729171753,
"learning_rate": 5e-05,
"loss": 1.9981966018676758,
"step": 2526
},
{
"epoch": 0.8814965915232665,
"grad_norm": 0.2508905529975891,
"learning_rate": 5e-05,
"loss": 1.8920555114746094,
"step": 2528
},
{
"epoch": 0.8821939780671932,
"grad_norm": 0.18876078724861145,
"learning_rate": 5e-05,
"loss": 1.944885492324829,
"step": 2530
},
{
"epoch": 0.8828913646111198,
"grad_norm": 0.20604079961776733,
"learning_rate": 5e-05,
"loss": 1.6713067293167114,
"step": 2532
},
{
"epoch": 0.8835887511550464,
"grad_norm": 0.21327614784240723,
"learning_rate": 5e-05,
"loss": 1.9312922954559326,
"step": 2534
},
{
"epoch": 0.8842861376989731,
"grad_norm": 0.24845241010189056,
"learning_rate": 5e-05,
"loss": 1.7195186614990234,
"step": 2536
},
{
"epoch": 0.8849835242428997,
"grad_norm": 0.2645180821418762,
"learning_rate": 5e-05,
"loss": 1.8852179050445557,
"step": 2538
},
{
"epoch": 0.8856809107868263,
"grad_norm": 0.16576892137527466,
"learning_rate": 5e-05,
"loss": 2.1656389236450195,
"step": 2540
},
{
"epoch": 0.886378297330753,
"grad_norm": 0.3268888592720032,
"learning_rate": 5e-05,
"loss": 1.7451763153076172,
"step": 2542
},
{
"epoch": 0.8870756838746796,
"grad_norm": 0.2656945586204529,
"learning_rate": 5e-05,
"loss": 1.654427409172058,
"step": 2544
},
{
"epoch": 0.8877730704186063,
"grad_norm": 0.20410767197608948,
"learning_rate": 5e-05,
"loss": 1.9891842603683472,
"step": 2546
},
{
"epoch": 0.8884704569625329,
"grad_norm": 0.17035138607025146,
"learning_rate": 5e-05,
"loss": 1.9926782846450806,
"step": 2548
},
{
"epoch": 0.8891678435064595,
"grad_norm": 0.37120959162712097,
"learning_rate": 5e-05,
"loss": 1.6961153745651245,
"step": 2550
},
{
"epoch": 0.8898652300503862,
"grad_norm": 0.22341012954711914,
"learning_rate": 5e-05,
"loss": 1.5798993110656738,
"step": 2552
},
{
"epoch": 0.8905626165943128,
"grad_norm": 0.236846923828125,
"learning_rate": 5e-05,
"loss": 2.0399224758148193,
"step": 2554
},
{
"epoch": 0.8912600031382395,
"grad_norm": 0.1636209487915039,
"learning_rate": 5e-05,
"loss": 1.870794415473938,
"step": 2556
},
{
"epoch": 0.8919573896821661,
"grad_norm": 0.22900991141796112,
"learning_rate": 5e-05,
"loss": 1.932793140411377,
"step": 2558
},
{
"epoch": 0.8926547762260927,
"grad_norm": 0.28880369663238525,
"learning_rate": 5e-05,
"loss": 1.7882640361785889,
"step": 2560
},
{
"epoch": 0.8933521627700194,
"grad_norm": 0.19175614416599274,
"learning_rate": 5e-05,
"loss": 1.9550821781158447,
"step": 2562
},
{
"epoch": 0.894049549313946,
"grad_norm": 0.23375819623470306,
"learning_rate": 5e-05,
"loss": 1.7253637313842773,
"step": 2564
},
{
"epoch": 0.8947469358578726,
"grad_norm": 0.27817198634147644,
"learning_rate": 5e-05,
"loss": 1.9125609397888184,
"step": 2566
},
{
"epoch": 0.8954443224017993,
"grad_norm": 0.21705344319343567,
"learning_rate": 5e-05,
"loss": 1.8138537406921387,
"step": 2568
},
{
"epoch": 0.8961417089457259,
"grad_norm": 0.3142533600330353,
"learning_rate": 5e-05,
"loss": 1.692766547203064,
"step": 2570
},
{
"epoch": 0.8968390954896526,
"grad_norm": 0.22375591099262238,
"learning_rate": 5e-05,
"loss": 1.9875195026397705,
"step": 2572
},
{
"epoch": 0.8975364820335792,
"grad_norm": 0.18041948974132538,
"learning_rate": 5e-05,
"loss": 1.9115439653396606,
"step": 2574
},
{
"epoch": 0.8982338685775058,
"grad_norm": 0.18039172887802124,
"learning_rate": 5e-05,
"loss": 1.860901117324829,
"step": 2576
},
{
"epoch": 0.8989312551214325,
"grad_norm": 0.2053157389163971,
"learning_rate": 5e-05,
"loss": 1.9434962272644043,
"step": 2578
},
{
"epoch": 0.899628641665359,
"grad_norm": 0.1571952998638153,
"learning_rate": 5e-05,
"loss": 1.936698079109192,
"step": 2580
},
{
"epoch": 0.9003260282092858,
"grad_norm": 0.21368089318275452,
"learning_rate": 5e-05,
"loss": 2.023897409439087,
"step": 2582
},
{
"epoch": 0.9010234147532123,
"grad_norm": 0.1864442527294159,
"learning_rate": 5e-05,
"loss": 2.1277942657470703,
"step": 2584
},
{
"epoch": 0.9017208012971389,
"grad_norm": 0.2126200795173645,
"learning_rate": 5e-05,
"loss": 1.7691402435302734,
"step": 2586
},
{
"epoch": 0.9024181878410656,
"grad_norm": 0.19210372865200043,
"learning_rate": 5e-05,
"loss": 1.9965739250183105,
"step": 2588
},
{
"epoch": 0.9031155743849922,
"grad_norm": 0.27291032671928406,
"learning_rate": 5e-05,
"loss": 1.896301031112671,
"step": 2590
},
{
"epoch": 0.9038129609289188,
"grad_norm": 0.24564699828624725,
"learning_rate": 5e-05,
"loss": 1.6648836135864258,
"step": 2592
},
{
"epoch": 0.9045103474728455,
"grad_norm": 0.2528457045555115,
"learning_rate": 5e-05,
"loss": 1.587568759918213,
"step": 2594
},
{
"epoch": 0.9052077340167721,
"grad_norm": 0.21464523673057556,
"learning_rate": 5e-05,
"loss": 1.9642722606658936,
"step": 2596
},
{
"epoch": 0.9059051205606988,
"grad_norm": 0.3087773621082306,
"learning_rate": 5e-05,
"loss": 1.8250300884246826,
"step": 2598
},
{
"epoch": 0.9066025071046254,
"grad_norm": 0.2622036337852478,
"learning_rate": 5e-05,
"loss": 1.8403692245483398,
"step": 2600
},
{
"epoch": 0.907299893648552,
"grad_norm": 0.19684091210365295,
"learning_rate": 5e-05,
"loss": 1.8768137693405151,
"step": 2602
},
{
"epoch": 0.9079972801924787,
"grad_norm": 0.23731729388237,
"learning_rate": 5e-05,
"loss": 1.6398066282272339,
"step": 2604
},
{
"epoch": 0.9086946667364053,
"grad_norm": 0.1656942069530487,
"learning_rate": 5e-05,
"loss": 2.0291547775268555,
"step": 2606
},
{
"epoch": 0.9093920532803319,
"grad_norm": 0.1387060135602951,
"learning_rate": 5e-05,
"loss": 1.8610889911651611,
"step": 2608
},
{
"epoch": 0.9100894398242586,
"grad_norm": 0.2080821543931961,
"learning_rate": 5e-05,
"loss": 2.0375406742095947,
"step": 2610
},
{
"epoch": 0.9107868263681852,
"grad_norm": 0.20191188156604767,
"learning_rate": 5e-05,
"loss": 1.9188122749328613,
"step": 2612
},
{
"epoch": 0.9114842129121119,
"grad_norm": 0.2041167914867401,
"learning_rate": 5e-05,
"loss": 1.6870101690292358,
"step": 2614
},
{
"epoch": 0.9121815994560385,
"grad_norm": 0.1676097959280014,
"learning_rate": 5e-05,
"loss": 2.0055911540985107,
"step": 2616
},
{
"epoch": 0.9128789859999651,
"grad_norm": 0.25617048144340515,
"learning_rate": 5e-05,
"loss": 1.8428096771240234,
"step": 2618
},
{
"epoch": 0.9135763725438918,
"grad_norm": 0.21877171099185944,
"learning_rate": 5e-05,
"loss": 1.909017562866211,
"step": 2620
},
{
"epoch": 0.9142737590878184,
"grad_norm": 0.22135400772094727,
"learning_rate": 5e-05,
"loss": 1.9755818843841553,
"step": 2622
},
{
"epoch": 0.9149711456317451,
"grad_norm": 0.26330921053886414,
"learning_rate": 5e-05,
"loss": 2.00614070892334,
"step": 2624
},
{
"epoch": 0.9156685321756717,
"grad_norm": 0.2363353669643402,
"learning_rate": 5e-05,
"loss": 1.9779236316680908,
"step": 2626
},
{
"epoch": 0.9163659187195983,
"grad_norm": 0.1923648715019226,
"learning_rate": 5e-05,
"loss": 2.1706905364990234,
"step": 2628
},
{
"epoch": 0.917063305263525,
"grad_norm": 0.2710999846458435,
"learning_rate": 5e-05,
"loss": 1.9036699533462524,
"step": 2630
},
{
"epoch": 0.9177606918074516,
"grad_norm": 0.22254317998886108,
"learning_rate": 5e-05,
"loss": 2.073990821838379,
"step": 2632
},
{
"epoch": 0.9184580783513782,
"grad_norm": 0.18711015582084656,
"learning_rate": 5e-05,
"loss": 1.9848815202713013,
"step": 2634
},
{
"epoch": 0.9191554648953049,
"grad_norm": 0.2281409651041031,
"learning_rate": 5e-05,
"loss": 1.8835316896438599,
"step": 2636
},
{
"epoch": 0.9198528514392315,
"grad_norm": 0.19834259152412415,
"learning_rate": 5e-05,
"loss": 1.7961277961730957,
"step": 2638
},
{
"epoch": 0.9205502379831582,
"grad_norm": 0.1919710338115692,
"learning_rate": 5e-05,
"loss": 1.780514121055603,
"step": 2640
},
{
"epoch": 0.9212476245270848,
"grad_norm": 0.20375385880470276,
"learning_rate": 5e-05,
"loss": 1.6803020238876343,
"step": 2642
},
{
"epoch": 0.9219450110710113,
"grad_norm": 0.17505325376987457,
"learning_rate": 5e-05,
"loss": 2.070007085800171,
"step": 2644
},
{
"epoch": 0.922642397614938,
"grad_norm": 0.21515130996704102,
"learning_rate": 5e-05,
"loss": 2.061142921447754,
"step": 2646
},
{
"epoch": 0.9233397841588646,
"grad_norm": 0.18638178706169128,
"learning_rate": 5e-05,
"loss": 1.846124529838562,
"step": 2648
},
{
"epoch": 0.9240371707027913,
"grad_norm": 0.22283639013767242,
"learning_rate": 5e-05,
"loss": 1.9895353317260742,
"step": 2650
},
{
"epoch": 0.9247345572467179,
"grad_norm": 0.18122327327728271,
"learning_rate": 5e-05,
"loss": 2.0273783206939697,
"step": 2652
},
{
"epoch": 0.9254319437906445,
"grad_norm": 0.1830211579799652,
"learning_rate": 5e-05,
"loss": 1.9389231204986572,
"step": 2654
},
{
"epoch": 0.9261293303345712,
"grad_norm": 0.208659827709198,
"learning_rate": 5e-05,
"loss": 1.9987062215805054,
"step": 2656
},
{
"epoch": 0.9268267168784978,
"grad_norm": 0.2385203242301941,
"learning_rate": 5e-05,
"loss": 1.8523343801498413,
"step": 2658
},
{
"epoch": 0.9275241034224244,
"grad_norm": 0.2189442217350006,
"learning_rate": 5e-05,
"loss": 1.9014790058135986,
"step": 2660
},
{
"epoch": 0.9282214899663511,
"grad_norm": 0.2747405171394348,
"learning_rate": 5e-05,
"loss": 1.873617172241211,
"step": 2662
},
{
"epoch": 0.9289188765102777,
"grad_norm": 0.24354961514472961,
"learning_rate": 5e-05,
"loss": 1.7986871004104614,
"step": 2664
},
{
"epoch": 0.9296162630542044,
"grad_norm": 0.2280634343624115,
"learning_rate": 5e-05,
"loss": 1.739278793334961,
"step": 2666
},
{
"epoch": 0.930313649598131,
"grad_norm": 0.3209410011768341,
"learning_rate": 5e-05,
"loss": 1.8125033378601074,
"step": 2668
},
{
"epoch": 0.9310110361420576,
"grad_norm": 0.2111290842294693,
"learning_rate": 5e-05,
"loss": 1.634263515472412,
"step": 2670
},
{
"epoch": 0.9317084226859843,
"grad_norm": 0.24231672286987305,
"learning_rate": 5e-05,
"loss": 1.8393033742904663,
"step": 2672
},
{
"epoch": 0.9324058092299109,
"grad_norm": 0.22962254285812378,
"learning_rate": 5e-05,
"loss": 1.6657109260559082,
"step": 2674
},
{
"epoch": 0.9331031957738375,
"grad_norm": 0.22991959750652313,
"learning_rate": 5e-05,
"loss": 1.7309353351593018,
"step": 2676
},
{
"epoch": 0.9338005823177642,
"grad_norm": 0.20495624840259552,
"learning_rate": 5e-05,
"loss": 1.9219999313354492,
"step": 2678
},
{
"epoch": 0.9344979688616908,
"grad_norm": 0.21537181735038757,
"learning_rate": 5e-05,
"loss": 1.7155791521072388,
"step": 2680
},
{
"epoch": 0.9351953554056175,
"grad_norm": 0.16606444120407104,
"learning_rate": 5e-05,
"loss": 1.9162561893463135,
"step": 2682
},
{
"epoch": 0.9358927419495441,
"grad_norm": 0.22539839148521423,
"learning_rate": 5e-05,
"loss": 2.145418167114258,
"step": 2684
},
{
"epoch": 0.9365901284934707,
"grad_norm": 0.21142470836639404,
"learning_rate": 5e-05,
"loss": 1.837956428527832,
"step": 2686
},
{
"epoch": 0.9372875150373974,
"grad_norm": 0.30611351132392883,
"learning_rate": 5e-05,
"loss": 1.99916672706604,
"step": 2688
},
{
"epoch": 0.937984901581324,
"grad_norm": 0.2546582520008087,
"learning_rate": 5e-05,
"loss": 1.8070523738861084,
"step": 2690
},
{
"epoch": 0.9386822881252507,
"grad_norm": 0.2777990698814392,
"learning_rate": 5e-05,
"loss": 1.8161311149597168,
"step": 2692
},
{
"epoch": 0.9393796746691773,
"grad_norm": 0.20838379859924316,
"learning_rate": 5e-05,
"loss": 1.7788996696472168,
"step": 2694
},
{
"epoch": 0.9400770612131039,
"grad_norm": 0.23629185557365417,
"learning_rate": 5e-05,
"loss": 1.9141902923583984,
"step": 2696
},
{
"epoch": 0.9407744477570306,
"grad_norm": 0.24336197972297668,
"learning_rate": 5e-05,
"loss": 1.840604543685913,
"step": 2698
},
{
"epoch": 0.9414718343009572,
"grad_norm": 0.23507782816886902,
"learning_rate": 5e-05,
"loss": 1.926959753036499,
"step": 2700
},
{
"epoch": 0.9421692208448837,
"grad_norm": 0.19096162915229797,
"learning_rate": 5e-05,
"loss": 1.8198974132537842,
"step": 2702
},
{
"epoch": 0.9428666073888105,
"grad_norm": 0.2538565397262573,
"learning_rate": 5e-05,
"loss": 1.8042070865631104,
"step": 2704
},
{
"epoch": 0.943563993932737,
"grad_norm": 0.27375662326812744,
"learning_rate": 5e-05,
"loss": 1.772554874420166,
"step": 2706
},
{
"epoch": 0.9442613804766637,
"grad_norm": 0.22935160994529724,
"learning_rate": 5e-05,
"loss": 1.8237781524658203,
"step": 2708
},
{
"epoch": 0.9449587670205903,
"grad_norm": 0.2059555947780609,
"learning_rate": 5e-05,
"loss": 1.8149011135101318,
"step": 2710
},
{
"epoch": 0.9456561535645169,
"grad_norm": 0.19966955482959747,
"learning_rate": 5e-05,
"loss": 1.8101370334625244,
"step": 2712
},
{
"epoch": 0.9463535401084436,
"grad_norm": 0.3353351354598999,
"learning_rate": 5e-05,
"loss": 1.7129199504852295,
"step": 2714
},
{
"epoch": 0.9470509266523702,
"grad_norm": 0.19486363232135773,
"learning_rate": 5e-05,
"loss": 2.043515205383301,
"step": 2716
},
{
"epoch": 0.9477483131962969,
"grad_norm": 0.22511737048625946,
"learning_rate": 5e-05,
"loss": 1.916710615158081,
"step": 2718
},
{
"epoch": 0.9484456997402235,
"grad_norm": 0.27413415908813477,
"learning_rate": 5e-05,
"loss": 1.7739595174789429,
"step": 2720
},
{
"epoch": 0.9491430862841501,
"grad_norm": 0.2778632342815399,
"learning_rate": 5e-05,
"loss": 1.6664798259735107,
"step": 2722
},
{
"epoch": 0.9498404728280768,
"grad_norm": 0.24520686268806458,
"learning_rate": 5e-05,
"loss": 1.8128085136413574,
"step": 2724
},
{
"epoch": 0.9505378593720034,
"grad_norm": 0.22817257046699524,
"learning_rate": 5e-05,
"loss": 1.9286627769470215,
"step": 2726
},
{
"epoch": 0.95123524591593,
"grad_norm": 0.22017443180084229,
"learning_rate": 5e-05,
"loss": 1.917222023010254,
"step": 2728
},
{
"epoch": 0.9519326324598567,
"grad_norm": 0.2212352454662323,
"learning_rate": 5e-05,
"loss": 1.952441692352295,
"step": 2730
},
{
"epoch": 0.9526300190037833,
"grad_norm": 0.23949624598026276,
"learning_rate": 5e-05,
"loss": 1.938012719154358,
"step": 2732
},
{
"epoch": 0.95332740554771,
"grad_norm": 0.19075579941272736,
"learning_rate": 5e-05,
"loss": 1.8858611583709717,
"step": 2734
},
{
"epoch": 0.9540247920916366,
"grad_norm": 0.19997099041938782,
"learning_rate": 5e-05,
"loss": 2.013164520263672,
"step": 2736
},
{
"epoch": 0.9547221786355632,
"grad_norm": 0.19334496557712555,
"learning_rate": 5e-05,
"loss": 1.986656904220581,
"step": 2738
},
{
"epoch": 0.9554195651794899,
"grad_norm": 0.2309669405221939,
"learning_rate": 5e-05,
"loss": 1.9071218967437744,
"step": 2740
},
{
"epoch": 0.9561169517234165,
"grad_norm": 0.26310646533966064,
"learning_rate": 5e-05,
"loss": 1.9484144449234009,
"step": 2742
},
{
"epoch": 0.9568143382673431,
"grad_norm": 0.15794895589351654,
"learning_rate": 5e-05,
"loss": 2.0087013244628906,
"step": 2744
},
{
"epoch": 0.9575117248112698,
"grad_norm": 0.2202284038066864,
"learning_rate": 5e-05,
"loss": 1.9209932088851929,
"step": 2746
},
{
"epoch": 0.9582091113551964,
"grad_norm": 0.22955642640590668,
"learning_rate": 5e-05,
"loss": 2.088130474090576,
"step": 2748
},
{
"epoch": 0.9589064978991231,
"grad_norm": 0.2963087558746338,
"learning_rate": 5e-05,
"loss": 1.8239340782165527,
"step": 2750
},
{
"epoch": 0.9596038844430497,
"grad_norm": 0.20487681031227112,
"learning_rate": 5e-05,
"loss": 2.0583481788635254,
"step": 2752
},
{
"epoch": 0.9603012709869763,
"grad_norm": 0.20250464975833893,
"learning_rate": 5e-05,
"loss": 1.787645936012268,
"step": 2754
},
{
"epoch": 0.960998657530903,
"grad_norm": 0.27794164419174194,
"learning_rate": 5e-05,
"loss": 1.8390411138534546,
"step": 2756
},
{
"epoch": 0.9616960440748296,
"grad_norm": 0.2529752254486084,
"learning_rate": 5e-05,
"loss": 2.0033533573150635,
"step": 2758
},
{
"epoch": 0.9623934306187563,
"grad_norm": 0.28755611181259155,
"learning_rate": 5e-05,
"loss": 2.0316929817199707,
"step": 2760
},
{
"epoch": 0.9630908171626829,
"grad_norm": 0.2371065616607666,
"learning_rate": 5e-05,
"loss": 1.9719016551971436,
"step": 2762
},
{
"epoch": 0.9637882037066094,
"grad_norm": 0.22012600302696228,
"learning_rate": 5e-05,
"loss": 2.0681281089782715,
"step": 2764
},
{
"epoch": 0.9644855902505362,
"grad_norm": 0.2199879139661789,
"learning_rate": 5e-05,
"loss": 1.8493678569793701,
"step": 2766
},
{
"epoch": 0.9651829767944627,
"grad_norm": 0.24364908039569855,
"learning_rate": 5e-05,
"loss": 1.8051460981369019,
"step": 2768
},
{
"epoch": 0.9658803633383893,
"grad_norm": 0.24932783842086792,
"learning_rate": 5e-05,
"loss": 1.8866969347000122,
"step": 2770
},
{
"epoch": 0.966577749882316,
"grad_norm": 0.2716982066631317,
"learning_rate": 5e-05,
"loss": 1.857971429824829,
"step": 2772
},
{
"epoch": 0.9672751364262426,
"grad_norm": 0.2361876666545868,
"learning_rate": 5e-05,
"loss": 1.664730429649353,
"step": 2774
},
{
"epoch": 0.9679725229701693,
"grad_norm": 0.17890170216560364,
"learning_rate": 5e-05,
"loss": 2.040210723876953,
"step": 2776
},
{
"epoch": 0.9686699095140959,
"grad_norm": 0.3918740451335907,
"learning_rate": 5e-05,
"loss": 1.542923927307129,
"step": 2778
},
{
"epoch": 0.9693672960580225,
"grad_norm": 0.2373792678117752,
"learning_rate": 5e-05,
"loss": 1.9707016944885254,
"step": 2780
},
{
"epoch": 0.9700646826019492,
"grad_norm": 0.18587234616279602,
"learning_rate": 5e-05,
"loss": 2.102854013442993,
"step": 2782
},
{
"epoch": 0.9707620691458758,
"grad_norm": 0.18375499546527863,
"learning_rate": 5e-05,
"loss": 2.152763843536377,
"step": 2784
},
{
"epoch": 0.9714594556898025,
"grad_norm": 0.17143887281417847,
"learning_rate": 5e-05,
"loss": 2.0487704277038574,
"step": 2786
},
{
"epoch": 0.9721568422337291,
"grad_norm": 0.2904473841190338,
"learning_rate": 5e-05,
"loss": 1.7972146272659302,
"step": 2788
},
{
"epoch": 0.9728542287776557,
"grad_norm": 0.20143947005271912,
"learning_rate": 5e-05,
"loss": 1.8335665464401245,
"step": 2790
},
{
"epoch": 0.9735516153215824,
"grad_norm": 0.2243218719959259,
"learning_rate": 5e-05,
"loss": 1.8498470783233643,
"step": 2792
},
{
"epoch": 0.974249001865509,
"grad_norm": 0.24106721580028534,
"learning_rate": 5e-05,
"loss": 1.962310552597046,
"step": 2794
},
{
"epoch": 0.9749463884094356,
"grad_norm": 0.19191642105579376,
"learning_rate": 5e-05,
"loss": 2.0525314807891846,
"step": 2796
},
{
"epoch": 0.9756437749533623,
"grad_norm": 0.25365757942199707,
"learning_rate": 5e-05,
"loss": 1.6377971172332764,
"step": 2798
},
{
"epoch": 0.9763411614972889,
"grad_norm": 0.30278539657592773,
"learning_rate": 5e-05,
"loss": 1.7661526203155518,
"step": 2800
},
{
"epoch": 0.9770385480412156,
"grad_norm": 0.19323121011257172,
"learning_rate": 5e-05,
"loss": 1.998081922531128,
"step": 2802
},
{
"epoch": 0.9777359345851422,
"grad_norm": 0.29371944069862366,
"learning_rate": 5e-05,
"loss": 2.128218412399292,
"step": 2804
},
{
"epoch": 0.9784333211290688,
"grad_norm": 0.2559215724468231,
"learning_rate": 5e-05,
"loss": 1.8026628494262695,
"step": 2806
},
{
"epoch": 0.9791307076729955,
"grad_norm": 0.2080441266298294,
"learning_rate": 5e-05,
"loss": 1.6901484727859497,
"step": 2808
},
{
"epoch": 0.9798280942169221,
"grad_norm": 0.24571523070335388,
"learning_rate": 5e-05,
"loss": 1.9744998216629028,
"step": 2810
},
{
"epoch": 0.9805254807608487,
"grad_norm": 0.17511089146137238,
"learning_rate": 5e-05,
"loss": 1.9440521001815796,
"step": 2812
},
{
"epoch": 0.9812228673047754,
"grad_norm": 0.1670263260602951,
"learning_rate": 5e-05,
"loss": 1.9503123760223389,
"step": 2814
},
{
"epoch": 0.981920253848702,
"grad_norm": 0.2044575959444046,
"learning_rate": 5e-05,
"loss": 1.8517839908599854,
"step": 2816
},
{
"epoch": 0.9826176403926287,
"grad_norm": 0.23012447357177734,
"learning_rate": 5e-05,
"loss": 1.9848601818084717,
"step": 2818
},
{
"epoch": 0.9833150269365553,
"grad_norm": 0.2421187460422516,
"learning_rate": 5e-05,
"loss": 1.8418478965759277,
"step": 2820
},
{
"epoch": 0.9840124134804819,
"grad_norm": 0.23758035898208618,
"learning_rate": 5e-05,
"loss": 1.9044866561889648,
"step": 2822
},
{
"epoch": 0.9847098000244086,
"grad_norm": 0.2850770056247711,
"learning_rate": 5e-05,
"loss": 1.841456413269043,
"step": 2824
},
{
"epoch": 0.9854071865683351,
"grad_norm": 0.15733706951141357,
"learning_rate": 5e-05,
"loss": 2.027836322784424,
"step": 2826
},
{
"epoch": 0.9861045731122619,
"grad_norm": 0.1889045536518097,
"learning_rate": 5e-05,
"loss": 1.814117193222046,
"step": 2828
},
{
"epoch": 0.9868019596561884,
"grad_norm": 0.19978995621204376,
"learning_rate": 5e-05,
"loss": 1.9840281009674072,
"step": 2830
},
{
"epoch": 0.987499346200115,
"grad_norm": 0.2702449858188629,
"learning_rate": 5e-05,
"loss": 1.6669400930404663,
"step": 2832
},
{
"epoch": 0.9881967327440417,
"grad_norm": 0.2253396213054657,
"learning_rate": 5e-05,
"loss": 1.9446139335632324,
"step": 2834
},
{
"epoch": 0.9888941192879683,
"grad_norm": 0.224853053689003,
"learning_rate": 5e-05,
"loss": 1.905352234840393,
"step": 2836
},
{
"epoch": 0.9895915058318949,
"grad_norm": 0.19716063141822815,
"learning_rate": 5e-05,
"loss": 1.749011516571045,
"step": 2838
},
{
"epoch": 0.9902888923758216,
"grad_norm": 0.21353597939014435,
"learning_rate": 5e-05,
"loss": 1.9643821716308594,
"step": 2840
},
{
"epoch": 0.9909862789197482,
"grad_norm": 0.2148216813802719,
"learning_rate": 5e-05,
"loss": 1.7480381727218628,
"step": 2842
},
{
"epoch": 0.9916836654636749,
"grad_norm": 0.23505060374736786,
"learning_rate": 5e-05,
"loss": 1.9068567752838135,
"step": 2844
},
{
"epoch": 0.9923810520076015,
"grad_norm": 0.16866576671600342,
"learning_rate": 5e-05,
"loss": 2.2292630672454834,
"step": 2846
},
{
"epoch": 0.9930784385515281,
"grad_norm": 0.3207954168319702,
"learning_rate": 5e-05,
"loss": 1.7519619464874268,
"step": 2848
},
{
"epoch": 0.9937758250954548,
"grad_norm": 0.21400299668312073,
"learning_rate": 5e-05,
"loss": 1.864077091217041,
"step": 2850
},
{
"epoch": 0.9944732116393814,
"grad_norm": 0.1865902990102768,
"learning_rate": 5e-05,
"loss": 2.105564832687378,
"step": 2852
},
{
"epoch": 0.9951705981833081,
"grad_norm": 0.18777675926685333,
"learning_rate": 5e-05,
"loss": 1.8181391954421997,
"step": 2854
},
{
"epoch": 0.9958679847272347,
"grad_norm": 0.22146295011043549,
"learning_rate": 5e-05,
"loss": 1.728520393371582,
"step": 2856
},
{
"epoch": 0.9965653712711613,
"grad_norm": 0.2045009434223175,
"learning_rate": 5e-05,
"loss": 2.065898895263672,
"step": 2858
},
{
"epoch": 0.997262757815088,
"grad_norm": 0.32111960649490356,
"learning_rate": 5e-05,
"loss": 1.8026396036148071,
"step": 2860
},
{
"epoch": 0.9979601443590146,
"grad_norm": 0.20345966517925262,
"learning_rate": 5e-05,
"loss": 2.041215419769287,
"step": 2862
},
{
"epoch": 0.9986575309029412,
"grad_norm": 0.22954697906970978,
"learning_rate": 5e-05,
"loss": 1.9399595260620117,
"step": 2864
},
{
"epoch": 0.9993549174468679,
"grad_norm": 0.2561749815940857,
"learning_rate": 5e-05,
"loss": 1.9872870445251465,
"step": 2866
},
{
"epoch": 1.0,
"grad_norm": 0.2276153415441513,
"learning_rate": 5e-05,
"loss": 1.9508090019226074,
"step": 2868
},
{
"epoch": 1.0006973865439266,
"grad_norm": 0.17299285531044006,
"learning_rate": 5e-05,
"loss": 1.8651812076568604,
"step": 2870
},
{
"epoch": 1.0013947730878532,
"grad_norm": 0.31359073519706726,
"learning_rate": 5e-05,
"loss": 1.829200029373169,
"step": 2872
},
{
"epoch": 1.00209215963178,
"grad_norm": 0.18494483828544617,
"learning_rate": 5e-05,
"loss": 1.7781418561935425,
"step": 2874
},
{
"epoch": 1.0027895461757066,
"grad_norm": 0.21310435235500336,
"learning_rate": 5e-05,
"loss": 1.9945908784866333,
"step": 2876
},
{
"epoch": 1.0034869327196332,
"grad_norm": 0.35635262727737427,
"learning_rate": 5e-05,
"loss": 1.9545801877975464,
"step": 2878
},
{
"epoch": 1.0041843192635598,
"grad_norm": 0.15053369104862213,
"learning_rate": 5e-05,
"loss": 1.9538103342056274,
"step": 2880
},
{
"epoch": 1.0048817058074864,
"grad_norm": 0.1651700884103775,
"learning_rate": 5e-05,
"loss": 1.8527746200561523,
"step": 2882
},
{
"epoch": 1.0055790923514132,
"grad_norm": 0.1976553052663803,
"learning_rate": 5e-05,
"loss": 1.9116666316986084,
"step": 2884
},
{
"epoch": 1.0062764788953398,
"grad_norm": 0.19889503717422485,
"learning_rate": 5e-05,
"loss": 1.7965984344482422,
"step": 2886
},
{
"epoch": 1.0069738654392664,
"grad_norm": 0.2198513001203537,
"learning_rate": 5e-05,
"loss": 1.806004285812378,
"step": 2888
},
{
"epoch": 1.007671251983193,
"grad_norm": 0.2889701724052429,
"learning_rate": 5e-05,
"loss": 1.7720134258270264,
"step": 2890
},
{
"epoch": 1.0083686385271196,
"grad_norm": 0.36210790276527405,
"learning_rate": 5e-05,
"loss": 1.713160514831543,
"step": 2892
},
{
"epoch": 1.0090660250710461,
"grad_norm": 0.276302307844162,
"learning_rate": 5e-05,
"loss": 1.9835500717163086,
"step": 2894
},
{
"epoch": 1.009763411614973,
"grad_norm": 0.22598609328269958,
"learning_rate": 5e-05,
"loss": 1.7208616733551025,
"step": 2896
},
{
"epoch": 1.0104607981588996,
"grad_norm": 0.24496674537658691,
"learning_rate": 5e-05,
"loss": 1.9628424644470215,
"step": 2898
},
{
"epoch": 1.0111581847028261,
"grad_norm": 0.2269727885723114,
"learning_rate": 5e-05,
"loss": 1.5986348390579224,
"step": 2900
},
{
"epoch": 1.0118555712467527,
"grad_norm": 0.1923527717590332,
"learning_rate": 5e-05,
"loss": 1.8378580808639526,
"step": 2902
},
{
"epoch": 1.0125529577906793,
"grad_norm": 0.28000184893608093,
"learning_rate": 5e-05,
"loss": 1.9745831489562988,
"step": 2904
},
{
"epoch": 1.0132503443346061,
"grad_norm": 0.1698903739452362,
"learning_rate": 5e-05,
"loss": 1.9972134828567505,
"step": 2906
},
{
"epoch": 1.0139477308785327,
"grad_norm": 0.24956777691841125,
"learning_rate": 5e-05,
"loss": 1.7017366886138916,
"step": 2908
},
{
"epoch": 1.0146451174224593,
"grad_norm": 0.2672395408153534,
"learning_rate": 5e-05,
"loss": 1.9146156311035156,
"step": 2910
},
{
"epoch": 1.015342503966386,
"grad_norm": 0.2607141435146332,
"learning_rate": 5e-05,
"loss": 1.779744267463684,
"step": 2912
},
{
"epoch": 1.0160398905103125,
"grad_norm": 0.24102437496185303,
"learning_rate": 5e-05,
"loss": 2.057152271270752,
"step": 2914
},
{
"epoch": 1.0167372770542393,
"grad_norm": 0.27756503224372864,
"learning_rate": 5e-05,
"loss": 1.8224923610687256,
"step": 2916
},
{
"epoch": 1.017434663598166,
"grad_norm": 0.22120578587055206,
"learning_rate": 5e-05,
"loss": 1.9365962743759155,
"step": 2918
},
{
"epoch": 1.0181320501420925,
"grad_norm": 0.22153885662555695,
"learning_rate": 5e-05,
"loss": 2.0323996543884277,
"step": 2920
},
{
"epoch": 1.018829436686019,
"grad_norm": 0.20578232407569885,
"learning_rate": 5e-05,
"loss": 1.768959403038025,
"step": 2922
},
{
"epoch": 1.0195268232299457,
"grad_norm": 0.18166182935237885,
"learning_rate": 5e-05,
"loss": 1.9475005865097046,
"step": 2924
},
{
"epoch": 1.0202242097738725,
"grad_norm": 0.21352547407150269,
"learning_rate": 5e-05,
"loss": 1.873194932937622,
"step": 2926
},
{
"epoch": 1.020921596317799,
"grad_norm": 0.2540004551410675,
"learning_rate": 5e-05,
"loss": 1.8963227272033691,
"step": 2928
},
{
"epoch": 1.0216189828617257,
"grad_norm": 0.22230134904384613,
"learning_rate": 5e-05,
"loss": 1.8723595142364502,
"step": 2930
},
{
"epoch": 1.0223163694056523,
"grad_norm": 0.20390909910202026,
"learning_rate": 5e-05,
"loss": 1.8132610321044922,
"step": 2932
},
{
"epoch": 1.0230137559495789,
"grad_norm": 0.3673308491706848,
"learning_rate": 5e-05,
"loss": 1.8942395448684692,
"step": 2934
},
{
"epoch": 1.0237111424935055,
"grad_norm": 0.2985556125640869,
"learning_rate": 5e-05,
"loss": 1.9247252941131592,
"step": 2936
},
{
"epoch": 1.0244085290374323,
"grad_norm": 0.21427017450332642,
"learning_rate": 5e-05,
"loss": 1.9653093814849854,
"step": 2938
},
{
"epoch": 1.0251059155813589,
"grad_norm": 0.22396492958068848,
"learning_rate": 5e-05,
"loss": 1.9198822975158691,
"step": 2940
},
{
"epoch": 1.0258033021252855,
"grad_norm": 0.18185977637767792,
"learning_rate": 5e-05,
"loss": 1.8885912895202637,
"step": 2942
},
{
"epoch": 1.026500688669212,
"grad_norm": 0.27587154507637024,
"learning_rate": 5e-05,
"loss": 2.0981240272521973,
"step": 2944
},
{
"epoch": 1.0271980752131387,
"grad_norm": 0.19659516215324402,
"learning_rate": 5e-05,
"loss": 2.030761241912842,
"step": 2946
},
{
"epoch": 1.0278954617570655,
"grad_norm": 0.2639068067073822,
"learning_rate": 5e-05,
"loss": 1.906769037246704,
"step": 2948
},
{
"epoch": 1.028592848300992,
"grad_norm": 0.19530723989009857,
"learning_rate": 5e-05,
"loss": 1.8872430324554443,
"step": 2950
},
{
"epoch": 1.0292902348449187,
"grad_norm": 0.3218066096305847,
"learning_rate": 5e-05,
"loss": 1.8188852071762085,
"step": 2952
},
{
"epoch": 1.0299876213888453,
"grad_norm": 0.22945411503314972,
"learning_rate": 5e-05,
"loss": 1.809032917022705,
"step": 2954
},
{
"epoch": 1.0306850079327718,
"grad_norm": 0.22216176986694336,
"learning_rate": 5e-05,
"loss": 1.8903319835662842,
"step": 2956
},
{
"epoch": 1.0313823944766987,
"grad_norm": 0.21097083389759064,
"learning_rate": 5e-05,
"loss": 2.15557861328125,
"step": 2958
},
{
"epoch": 1.0320797810206253,
"grad_norm": 0.16663368046283722,
"learning_rate": 5e-05,
"loss": 1.778777837753296,
"step": 2960
},
{
"epoch": 1.0327771675645518,
"grad_norm": 0.27202552556991577,
"learning_rate": 5e-05,
"loss": 1.908647060394287,
"step": 2962
},
{
"epoch": 1.0334745541084784,
"grad_norm": 0.18464124202728271,
"learning_rate": 5e-05,
"loss": 1.9011168479919434,
"step": 2964
},
{
"epoch": 1.034171940652405,
"grad_norm": 0.2257303148508072,
"learning_rate": 5e-05,
"loss": 1.8555386066436768,
"step": 2966
},
{
"epoch": 1.0348693271963318,
"grad_norm": 0.3158121109008789,
"learning_rate": 5e-05,
"loss": 1.880272626876831,
"step": 2968
},
{
"epoch": 1.0355667137402584,
"grad_norm": 0.15654286742210388,
"learning_rate": 5e-05,
"loss": 1.8980830907821655,
"step": 2970
},
{
"epoch": 1.036264100284185,
"grad_norm": 0.20344379544258118,
"learning_rate": 5e-05,
"loss": 1.9919698238372803,
"step": 2972
},
{
"epoch": 1.0369614868281116,
"grad_norm": 0.21301423013210297,
"learning_rate": 5e-05,
"loss": 1.8638865947723389,
"step": 2974
},
{
"epoch": 1.0376588733720382,
"grad_norm": 0.17934224009513855,
"learning_rate": 5e-05,
"loss": 1.9926419258117676,
"step": 2976
},
{
"epoch": 1.038356259915965,
"grad_norm": 0.19874687492847443,
"learning_rate": 5e-05,
"loss": 1.969261884689331,
"step": 2978
},
{
"epoch": 1.0390536464598916,
"grad_norm": 0.23959201574325562,
"learning_rate": 5e-05,
"loss": 1.9629769325256348,
"step": 2980
},
{
"epoch": 1.0397510330038182,
"grad_norm": 0.18820637464523315,
"learning_rate": 5e-05,
"loss": 1.990842580795288,
"step": 2982
},
{
"epoch": 1.0404484195477448,
"grad_norm": 0.3176293671131134,
"learning_rate": 5e-05,
"loss": 1.7065595388412476,
"step": 2984
},
{
"epoch": 1.0411458060916714,
"grad_norm": 0.3154110312461853,
"learning_rate": 5e-05,
"loss": 1.879931926727295,
"step": 2986
},
{
"epoch": 1.041843192635598,
"grad_norm": 0.18767651915550232,
"learning_rate": 5e-05,
"loss": 1.9425246715545654,
"step": 2988
},
{
"epoch": 1.0425405791795248,
"grad_norm": 0.42967626452445984,
"learning_rate": 5e-05,
"loss": 1.9072425365447998,
"step": 2990
},
{
"epoch": 1.0432379657234514,
"grad_norm": 0.17276839911937714,
"learning_rate": 5e-05,
"loss": 1.930345058441162,
"step": 2992
},
{
"epoch": 1.043935352267378,
"grad_norm": 0.20109118521213531,
"learning_rate": 5e-05,
"loss": 1.82950758934021,
"step": 2994
},
{
"epoch": 1.0446327388113046,
"grad_norm": 0.19736596941947937,
"learning_rate": 5e-05,
"loss": 2.118715763092041,
"step": 2996
},
{
"epoch": 1.0453301253552312,
"grad_norm": 0.20882341265678406,
"learning_rate": 5e-05,
"loss": 1.8056647777557373,
"step": 2998
},
{
"epoch": 1.046027511899158,
"grad_norm": 0.24335800111293793,
"learning_rate": 5e-05,
"loss": 2.065293073654175,
"step": 3000
},
{
"epoch": 1.0467248984430846,
"grad_norm": 0.20033885538578033,
"learning_rate": 5e-05,
"loss": 1.9977822303771973,
"step": 3002
},
{
"epoch": 1.0474222849870112,
"grad_norm": 0.23461225628852844,
"learning_rate": 5e-05,
"loss": 1.9835517406463623,
"step": 3004
},
{
"epoch": 1.0481196715309378,
"grad_norm": 0.18256652355194092,
"learning_rate": 5e-05,
"loss": 1.7411243915557861,
"step": 3006
},
{
"epoch": 1.0488170580748644,
"grad_norm": 0.18111863732337952,
"learning_rate": 5e-05,
"loss": 1.7137845754623413,
"step": 3008
},
{
"epoch": 1.0495144446187912,
"grad_norm": 0.3122173845767975,
"learning_rate": 5e-05,
"loss": 1.684222936630249,
"step": 3010
},
{
"epoch": 1.0502118311627178,
"grad_norm": 0.39914849400520325,
"learning_rate": 5e-05,
"loss": 1.5934462547302246,
"step": 3012
},
{
"epoch": 1.0509092177066444,
"grad_norm": 0.17329807579517365,
"learning_rate": 5e-05,
"loss": 1.8913609981536865,
"step": 3014
},
{
"epoch": 1.051606604250571,
"grad_norm": 0.24272692203521729,
"learning_rate": 5e-05,
"loss": 1.8258047103881836,
"step": 3016
},
{
"epoch": 1.0523039907944975,
"grad_norm": 0.20756541192531586,
"learning_rate": 5e-05,
"loss": 1.8831913471221924,
"step": 3018
},
{
"epoch": 1.0530013773384244,
"grad_norm": 0.24580636620521545,
"learning_rate": 5e-05,
"loss": 1.7874979972839355,
"step": 3020
},
{
"epoch": 1.053698763882351,
"grad_norm": 0.211761012673378,
"learning_rate": 5e-05,
"loss": 1.848203420639038,
"step": 3022
},
{
"epoch": 1.0543961504262775,
"grad_norm": 0.27407196164131165,
"learning_rate": 5e-05,
"loss": 1.7207894325256348,
"step": 3024
},
{
"epoch": 1.0550935369702041,
"grad_norm": 0.22823724150657654,
"learning_rate": 5e-05,
"loss": 1.9045991897583008,
"step": 3026
},
{
"epoch": 1.0557909235141307,
"grad_norm": 0.3093097507953644,
"learning_rate": 5e-05,
"loss": 1.8167511224746704,
"step": 3028
},
{
"epoch": 1.0564883100580573,
"grad_norm": 0.20107699930667877,
"learning_rate": 5e-05,
"loss": 1.9803643226623535,
"step": 3030
},
{
"epoch": 1.0571856966019841,
"grad_norm": 0.2670706808567047,
"learning_rate": 5e-05,
"loss": 1.766595482826233,
"step": 3032
},
{
"epoch": 1.0578830831459107,
"grad_norm": 0.265388548374176,
"learning_rate": 5e-05,
"loss": 1.7853655815124512,
"step": 3034
},
{
"epoch": 1.0585804696898373,
"grad_norm": 0.24559935927391052,
"learning_rate": 5e-05,
"loss": 1.942050576210022,
"step": 3036
},
{
"epoch": 1.059277856233764,
"grad_norm": 0.1944202333688736,
"learning_rate": 5e-05,
"loss": 1.7787821292877197,
"step": 3038
},
{
"epoch": 1.0599752427776905,
"grad_norm": 0.3455706536769867,
"learning_rate": 5e-05,
"loss": 1.726896047592163,
"step": 3040
},
{
"epoch": 1.0606726293216173,
"grad_norm": 0.19156025350093842,
"learning_rate": 5e-05,
"loss": 1.9844861030578613,
"step": 3042
},
{
"epoch": 1.061370015865544,
"grad_norm": 0.45266610383987427,
"learning_rate": 5e-05,
"loss": 1.8766891956329346,
"step": 3044
},
{
"epoch": 1.0620674024094705,
"grad_norm": 0.2506283223628998,
"learning_rate": 5e-05,
"loss": 2.065756320953369,
"step": 3046
},
{
"epoch": 1.062764788953397,
"grad_norm": 0.2267516851425171,
"learning_rate": 5e-05,
"loss": 1.7935112714767456,
"step": 3048
},
{
"epoch": 1.0634621754973237,
"grad_norm": 0.2037196010351181,
"learning_rate": 5e-05,
"loss": 1.9053187370300293,
"step": 3050
},
{
"epoch": 1.0641595620412505,
"grad_norm": 0.18256697058677673,
"learning_rate": 5e-05,
"loss": 1.9324783086776733,
"step": 3052
},
{
"epoch": 1.064856948585177,
"grad_norm": 0.29540106654167175,
"learning_rate": 5e-05,
"loss": 1.8924736976623535,
"step": 3054
},
{
"epoch": 1.0655543351291037,
"grad_norm": 0.222468763589859,
"learning_rate": 5e-05,
"loss": 1.962878942489624,
"step": 3056
},
{
"epoch": 1.0662517216730303,
"grad_norm": 0.2955968677997589,
"learning_rate": 5e-05,
"loss": 1.7855193614959717,
"step": 3058
},
{
"epoch": 1.0669491082169569,
"grad_norm": 0.22217248380184174,
"learning_rate": 5e-05,
"loss": 2.028803825378418,
"step": 3060
},
{
"epoch": 1.0676464947608837,
"grad_norm": 0.2518015503883362,
"learning_rate": 5e-05,
"loss": 1.8943895101547241,
"step": 3062
},
{
"epoch": 1.0683438813048103,
"grad_norm": 0.20429520308971405,
"learning_rate": 5e-05,
"loss": 1.837890386581421,
"step": 3064
},
{
"epoch": 1.0690412678487369,
"grad_norm": 0.28744643926620483,
"learning_rate": 5e-05,
"loss": 1.9460480213165283,
"step": 3066
},
{
"epoch": 1.0697386543926635,
"grad_norm": 0.19168125092983246,
"learning_rate": 5e-05,
"loss": 2.061906337738037,
"step": 3068
},
{
"epoch": 1.07043604093659,
"grad_norm": 0.19340719282627106,
"learning_rate": 5e-05,
"loss": 1.8375229835510254,
"step": 3070
},
{
"epoch": 1.0711334274805169,
"grad_norm": 0.17509004473686218,
"learning_rate": 5e-05,
"loss": 2.0236008167266846,
"step": 3072
},
{
"epoch": 1.0718308140244435,
"grad_norm": 0.2244594395160675,
"learning_rate": 5e-05,
"loss": 1.7777913808822632,
"step": 3074
},
{
"epoch": 1.07252820056837,
"grad_norm": 0.16912101209163666,
"learning_rate": 5e-05,
"loss": 1.7235589027404785,
"step": 3076
},
{
"epoch": 1.0732255871122967,
"grad_norm": 0.1829208880662918,
"learning_rate": 5e-05,
"loss": 1.9091994762420654,
"step": 3078
},
{
"epoch": 1.0739229736562232,
"grad_norm": 0.1684057116508484,
"learning_rate": 5e-05,
"loss": 2.081568479537964,
"step": 3080
},
{
"epoch": 1.0746203602001498,
"grad_norm": 0.18698105216026306,
"learning_rate": 5e-05,
"loss": 1.943549633026123,
"step": 3082
},
{
"epoch": 1.0753177467440767,
"grad_norm": 0.24537481367588043,
"learning_rate": 5e-05,
"loss": 1.8758091926574707,
"step": 3084
},
{
"epoch": 1.0760151332880032,
"grad_norm": 0.2622872591018677,
"learning_rate": 5e-05,
"loss": 1.7700133323669434,
"step": 3086
},
{
"epoch": 1.0767125198319298,
"grad_norm": 0.18435196578502655,
"learning_rate": 5e-05,
"loss": 2.1117703914642334,
"step": 3088
},
{
"epoch": 1.0774099063758564,
"grad_norm": 0.32609033584594727,
"learning_rate": 5e-05,
"loss": 1.747119426727295,
"step": 3090
},
{
"epoch": 1.078107292919783,
"grad_norm": 0.1847252994775772,
"learning_rate": 5e-05,
"loss": 1.8103517293930054,
"step": 3092
},
{
"epoch": 1.0788046794637098,
"grad_norm": 0.24559560418128967,
"learning_rate": 5e-05,
"loss": 1.8911972045898438,
"step": 3094
},
{
"epoch": 1.0795020660076364,
"grad_norm": 0.23133735358715057,
"learning_rate": 5e-05,
"loss": 2.0859274864196777,
"step": 3096
},
{
"epoch": 1.080199452551563,
"grad_norm": 0.3087204396724701,
"learning_rate": 5e-05,
"loss": 2.0171196460723877,
"step": 3098
},
{
"epoch": 1.0808968390954896,
"grad_norm": 0.1970721334218979,
"learning_rate": 5e-05,
"loss": 1.945678949356079,
"step": 3100
},
{
"epoch": 1.0815942256394162,
"grad_norm": 0.33208608627319336,
"learning_rate": 5e-05,
"loss": 1.496478796005249,
"step": 3102
},
{
"epoch": 1.082291612183343,
"grad_norm": 0.2139287292957306,
"learning_rate": 5e-05,
"loss": 1.9864192008972168,
"step": 3104
},
{
"epoch": 1.0829889987272696,
"grad_norm": 0.19848096370697021,
"learning_rate": 5e-05,
"loss": 1.962013602256775,
"step": 3106
},
{
"epoch": 1.0836863852711962,
"grad_norm": 0.27842608094215393,
"learning_rate": 5e-05,
"loss": 1.8546619415283203,
"step": 3108
},
{
"epoch": 1.0843837718151228,
"grad_norm": 0.2816014885902405,
"learning_rate": 5e-05,
"loss": 1.6673271656036377,
"step": 3110
},
{
"epoch": 1.0850811583590494,
"grad_norm": 0.18917010724544525,
"learning_rate": 5e-05,
"loss": 1.8602776527404785,
"step": 3112
},
{
"epoch": 1.085778544902976,
"grad_norm": 0.19316600263118744,
"learning_rate": 5e-05,
"loss": 1.9155161380767822,
"step": 3114
},
{
"epoch": 1.0864759314469028,
"grad_norm": 0.2505404055118561,
"learning_rate": 5e-05,
"loss": 1.8740239143371582,
"step": 3116
},
{
"epoch": 1.0871733179908294,
"grad_norm": 0.21722301840782166,
"learning_rate": 5e-05,
"loss": 1.8883038759231567,
"step": 3118
},
{
"epoch": 1.087870704534756,
"grad_norm": 0.3146092891693115,
"learning_rate": 5e-05,
"loss": 2.1066112518310547,
"step": 3120
},
{
"epoch": 1.0885680910786826,
"grad_norm": 0.1916327029466629,
"learning_rate": 5e-05,
"loss": 1.7462494373321533,
"step": 3122
},
{
"epoch": 1.0892654776226092,
"grad_norm": 0.23770633339881897,
"learning_rate": 5e-05,
"loss": 1.9702515602111816,
"step": 3124
},
{
"epoch": 1.089962864166536,
"grad_norm": 0.2103680968284607,
"learning_rate": 5e-05,
"loss": 1.9769155979156494,
"step": 3126
},
{
"epoch": 1.0906602507104626,
"grad_norm": 0.20509839057922363,
"learning_rate": 5e-05,
"loss": 1.915466070175171,
"step": 3128
},
{
"epoch": 1.0913576372543892,
"grad_norm": 0.20828916132450104,
"learning_rate": 5e-05,
"loss": 1.885035753250122,
"step": 3130
},
{
"epoch": 1.0920550237983158,
"grad_norm": 0.23557862639427185,
"learning_rate": 5e-05,
"loss": 2.0801403522491455,
"step": 3132
},
{
"epoch": 1.0927524103422424,
"grad_norm": 0.18909405171871185,
"learning_rate": 5e-05,
"loss": 2.022665023803711,
"step": 3134
},
{
"epoch": 1.0934497968861692,
"grad_norm": 0.15867860615253448,
"learning_rate": 5e-05,
"loss": 1.9656434059143066,
"step": 3136
},
{
"epoch": 1.0941471834300958,
"grad_norm": 0.18615524470806122,
"learning_rate": 5e-05,
"loss": 1.851471185684204,
"step": 3138
},
{
"epoch": 1.0948445699740224,
"grad_norm": 0.1827930510044098,
"learning_rate": 5e-05,
"loss": 1.811725378036499,
"step": 3140
},
{
"epoch": 1.095541956517949,
"grad_norm": 0.2323254495859146,
"learning_rate": 5e-05,
"loss": 1.9539238214492798,
"step": 3142
},
{
"epoch": 1.0962393430618755,
"grad_norm": 0.20112617313861847,
"learning_rate": 5e-05,
"loss": 1.9900753498077393,
"step": 3144
},
{
"epoch": 1.0969367296058024,
"grad_norm": 0.2772871255874634,
"learning_rate": 5e-05,
"loss": 1.8737435340881348,
"step": 3146
},
{
"epoch": 1.097634116149729,
"grad_norm": 0.2420884668827057,
"learning_rate": 5e-05,
"loss": 1.8864036798477173,
"step": 3148
},
{
"epoch": 1.0983315026936555,
"grad_norm": 0.22212693095207214,
"learning_rate": 5e-05,
"loss": 1.9165153503417969,
"step": 3150
},
{
"epoch": 1.0990288892375821,
"grad_norm": 0.17598916590213776,
"learning_rate": 5e-05,
"loss": 1.9989866018295288,
"step": 3152
},
{
"epoch": 1.0997262757815087,
"grad_norm": 0.14702627062797546,
"learning_rate": 5e-05,
"loss": 1.9490489959716797,
"step": 3154
},
{
"epoch": 1.1004236623254355,
"grad_norm": 0.19959218800067902,
"learning_rate": 5e-05,
"loss": 1.7468793392181396,
"step": 3156
},
{
"epoch": 1.1011210488693621,
"grad_norm": 0.19527176022529602,
"learning_rate": 5e-05,
"loss": 1.9752986431121826,
"step": 3158
},
{
"epoch": 1.1018184354132887,
"grad_norm": 0.15109111368656158,
"learning_rate": 5e-05,
"loss": 1.995666742324829,
"step": 3160
},
{
"epoch": 1.1025158219572153,
"grad_norm": 0.26133108139038086,
"learning_rate": 5e-05,
"loss": 1.740058422088623,
"step": 3162
},
{
"epoch": 1.103213208501142,
"grad_norm": 0.24775837361812592,
"learning_rate": 5e-05,
"loss": 1.9653204679489136,
"step": 3164
},
{
"epoch": 1.1039105950450687,
"grad_norm": 0.23073656857013702,
"learning_rate": 5e-05,
"loss": 1.8970496654510498,
"step": 3166
},
{
"epoch": 1.1046079815889953,
"grad_norm": 0.2658872902393341,
"learning_rate": 5e-05,
"loss": 1.9253250360488892,
"step": 3168
},
{
"epoch": 1.105305368132922,
"grad_norm": 0.17600391805171967,
"learning_rate": 5e-05,
"loss": 2.069902181625366,
"step": 3170
},
{
"epoch": 1.1060027546768485,
"grad_norm": 0.18817438185214996,
"learning_rate": 5e-05,
"loss": 2.0579793453216553,
"step": 3172
},
{
"epoch": 1.106700141220775,
"grad_norm": 0.2382892221212387,
"learning_rate": 5e-05,
"loss": 1.797858476638794,
"step": 3174
},
{
"epoch": 1.1073975277647017,
"grad_norm": 0.2064724713563919,
"learning_rate": 5e-05,
"loss": 1.8315250873565674,
"step": 3176
},
{
"epoch": 1.1080949143086285,
"grad_norm": 0.26791125535964966,
"learning_rate": 5e-05,
"loss": 1.9984941482543945,
"step": 3178
},
{
"epoch": 1.108792300852555,
"grad_norm": 0.20273837447166443,
"learning_rate": 5e-05,
"loss": 1.9348340034484863,
"step": 3180
},
{
"epoch": 1.1094896873964817,
"grad_norm": 0.21015384793281555,
"learning_rate": 5e-05,
"loss": 1.895909070968628,
"step": 3182
},
{
"epoch": 1.1101870739404083,
"grad_norm": 0.22125685214996338,
"learning_rate": 5e-05,
"loss": 2.190981149673462,
"step": 3184
},
{
"epoch": 1.1108844604843349,
"grad_norm": 0.2263241708278656,
"learning_rate": 5e-05,
"loss": 1.9944695234298706,
"step": 3186
},
{
"epoch": 1.1115818470282617,
"grad_norm": 0.27125614881515503,
"learning_rate": 5e-05,
"loss": 1.952336311340332,
"step": 3188
},
{
"epoch": 1.1122792335721883,
"grad_norm": 0.21309427917003632,
"learning_rate": 5e-05,
"loss": 1.8668360710144043,
"step": 3190
},
{
"epoch": 1.1129766201161149,
"grad_norm": 0.20125311613082886,
"learning_rate": 5e-05,
"loss": 1.8131139278411865,
"step": 3192
},
{
"epoch": 1.1136740066600415,
"grad_norm": 0.22964510321617126,
"learning_rate": 5e-05,
"loss": 1.6908190250396729,
"step": 3194
},
{
"epoch": 1.114371393203968,
"grad_norm": 0.2215379923582077,
"learning_rate": 5e-05,
"loss": 2.077676296234131,
"step": 3196
},
{
"epoch": 1.1150687797478949,
"grad_norm": 0.19939805567264557,
"learning_rate": 5e-05,
"loss": 1.8063565492630005,
"step": 3198
},
{
"epoch": 1.1157661662918215,
"grad_norm": 0.3067377805709839,
"learning_rate": 5e-05,
"loss": 1.9783754348754883,
"step": 3200
},
{
"epoch": 1.116463552835748,
"grad_norm": 0.33919110894203186,
"learning_rate": 5e-05,
"loss": 1.90299391746521,
"step": 3202
},
{
"epoch": 1.1171609393796746,
"grad_norm": 0.2887316942214966,
"learning_rate": 5e-05,
"loss": 1.856370210647583,
"step": 3204
},
{
"epoch": 1.1178583259236012,
"grad_norm": 0.24907241761684418,
"learning_rate": 5e-05,
"loss": 1.9177173376083374,
"step": 3206
},
{
"epoch": 1.1185557124675278,
"grad_norm": 0.23588360846042633,
"learning_rate": 5e-05,
"loss": 1.893583059310913,
"step": 3208
},
{
"epoch": 1.1192530990114546,
"grad_norm": 0.3276607096195221,
"learning_rate": 5e-05,
"loss": 1.797567367553711,
"step": 3210
},
{
"epoch": 1.1199504855553812,
"grad_norm": 0.24757027626037598,
"learning_rate": 5e-05,
"loss": 1.9606432914733887,
"step": 3212
},
{
"epoch": 1.1206478720993078,
"grad_norm": 0.23869870603084564,
"learning_rate": 5e-05,
"loss": 1.818904161453247,
"step": 3214
},
{
"epoch": 1.1213452586432344,
"grad_norm": 0.27114957571029663,
"learning_rate": 5e-05,
"loss": 1.8644049167633057,
"step": 3216
},
{
"epoch": 1.122042645187161,
"grad_norm": 0.17666642367839813,
"learning_rate": 5e-05,
"loss": 1.9567948579788208,
"step": 3218
},
{
"epoch": 1.1227400317310878,
"grad_norm": 0.22872596979141235,
"learning_rate": 5e-05,
"loss": 1.8192429542541504,
"step": 3220
},
{
"epoch": 1.1234374182750144,
"grad_norm": 0.17448724806308746,
"learning_rate": 5e-05,
"loss": 1.9359086751937866,
"step": 3222
},
{
"epoch": 1.124134804818941,
"grad_norm": 0.1999889314174652,
"learning_rate": 5e-05,
"loss": 2.0217783451080322,
"step": 3224
},
{
"epoch": 1.1248321913628676,
"grad_norm": 0.22751814126968384,
"learning_rate": 5e-05,
"loss": 1.941468358039856,
"step": 3226
},
{
"epoch": 1.1255295779067942,
"grad_norm": 0.18390969932079315,
"learning_rate": 5e-05,
"loss": 1.9881794452667236,
"step": 3228
},
{
"epoch": 1.126226964450721,
"grad_norm": 0.2620185911655426,
"learning_rate": 5e-05,
"loss": 1.9302098751068115,
"step": 3230
},
{
"epoch": 1.1269243509946476,
"grad_norm": 0.21205852925777435,
"learning_rate": 5e-05,
"loss": 1.7991564273834229,
"step": 3232
},
{
"epoch": 1.1276217375385742,
"grad_norm": 0.2107517570257187,
"learning_rate": 5e-05,
"loss": 1.7906372547149658,
"step": 3234
},
{
"epoch": 1.1283191240825008,
"grad_norm": 0.2992035746574402,
"learning_rate": 5e-05,
"loss": 1.7822850942611694,
"step": 3236
},
{
"epoch": 1.1290165106264274,
"grad_norm": 0.2746608555316925,
"learning_rate": 5e-05,
"loss": 1.8279621601104736,
"step": 3238
},
{
"epoch": 1.1297138971703542,
"grad_norm": 0.20095936954021454,
"learning_rate": 5e-05,
"loss": 2.232962131500244,
"step": 3240
},
{
"epoch": 1.1304112837142808,
"grad_norm": 0.17785291373729706,
"learning_rate": 5e-05,
"loss": 1.9122307300567627,
"step": 3242
},
{
"epoch": 1.1311086702582074,
"grad_norm": 0.21463751792907715,
"learning_rate": 5e-05,
"loss": 1.7621231079101562,
"step": 3244
},
{
"epoch": 1.131806056802134,
"grad_norm": 0.24938708543777466,
"learning_rate": 5e-05,
"loss": 1.876266598701477,
"step": 3246
},
{
"epoch": 1.1325034433460606,
"grad_norm": 0.17530113458633423,
"learning_rate": 5e-05,
"loss": 1.9535584449768066,
"step": 3248
},
{
"epoch": 1.1332008298899874,
"grad_norm": 0.211569681763649,
"learning_rate": 5e-05,
"loss": 1.862959384918213,
"step": 3250
},
{
"epoch": 1.133898216433914,
"grad_norm": 0.17921343445777893,
"learning_rate": 5e-05,
"loss": 1.8929667472839355,
"step": 3252
},
{
"epoch": 1.1345956029778406,
"grad_norm": 0.21480803191661835,
"learning_rate": 5e-05,
"loss": 1.775618076324463,
"step": 3254
},
{
"epoch": 1.1352929895217672,
"grad_norm": 0.1842929869890213,
"learning_rate": 5e-05,
"loss": 2.046151876449585,
"step": 3256
},
{
"epoch": 1.1359903760656938,
"grad_norm": 0.2910834550857544,
"learning_rate": 5e-05,
"loss": 1.8534057140350342,
"step": 3258
},
{
"epoch": 1.1366877626096206,
"grad_norm": 0.23228909075260162,
"learning_rate": 5e-05,
"loss": 2.0110678672790527,
"step": 3260
},
{
"epoch": 1.1373851491535472,
"grad_norm": 0.16914640367031097,
"learning_rate": 5e-05,
"loss": 1.95835280418396,
"step": 3262
},
{
"epoch": 1.1380825356974738,
"grad_norm": 0.2415015995502472,
"learning_rate": 5e-05,
"loss": 1.965360403060913,
"step": 3264
},
{
"epoch": 1.1387799222414003,
"grad_norm": 0.188995823264122,
"learning_rate": 5e-05,
"loss": 1.761688470840454,
"step": 3266
},
{
"epoch": 1.139477308785327,
"grad_norm": 0.27970463037490845,
"learning_rate": 5e-05,
"loss": 2.0222902297973633,
"step": 3268
},
{
"epoch": 1.1401746953292535,
"grad_norm": 0.26102328300476074,
"learning_rate": 5e-05,
"loss": 1.8212395906448364,
"step": 3270
},
{
"epoch": 1.1408720818731803,
"grad_norm": 0.218557208776474,
"learning_rate": 5e-05,
"loss": 1.816208839416504,
"step": 3272
},
{
"epoch": 1.141569468417107,
"grad_norm": 0.20457427203655243,
"learning_rate": 5e-05,
"loss": 1.786121129989624,
"step": 3274
},
{
"epoch": 1.1422668549610335,
"grad_norm": 0.2177436649799347,
"learning_rate": 5e-05,
"loss": 1.9937255382537842,
"step": 3276
},
{
"epoch": 1.1429642415049601,
"grad_norm": 0.24371139705181122,
"learning_rate": 5e-05,
"loss": 1.6575055122375488,
"step": 3278
},
{
"epoch": 1.1436616280488867,
"grad_norm": 0.22388990223407745,
"learning_rate": 5e-05,
"loss": 2.0424928665161133,
"step": 3280
},
{
"epoch": 1.1443590145928135,
"grad_norm": 0.15538032352924347,
"learning_rate": 5e-05,
"loss": 2.019500494003296,
"step": 3282
},
{
"epoch": 1.1450564011367401,
"grad_norm": 0.24666345119476318,
"learning_rate": 5e-05,
"loss": 1.904087781906128,
"step": 3284
},
{
"epoch": 1.1457537876806667,
"grad_norm": 0.2622423470020294,
"learning_rate": 5e-05,
"loss": 1.9055876731872559,
"step": 3286
},
{
"epoch": 1.1464511742245933,
"grad_norm": 0.2091299444437027,
"learning_rate": 5e-05,
"loss": 1.9555625915527344,
"step": 3288
},
{
"epoch": 1.14714856076852,
"grad_norm": 0.25040286779403687,
"learning_rate": 5e-05,
"loss": 1.7698583602905273,
"step": 3290
},
{
"epoch": 1.1478459473124465,
"grad_norm": 0.22865115106105804,
"learning_rate": 5e-05,
"loss": 1.8670485019683838,
"step": 3292
},
{
"epoch": 1.1485433338563733,
"grad_norm": 0.19878660142421722,
"learning_rate": 5e-05,
"loss": 1.9411699771881104,
"step": 3294
},
{
"epoch": 1.1492407204003,
"grad_norm": 0.16737036406993866,
"learning_rate": 5e-05,
"loss": 1.9980804920196533,
"step": 3296
},
{
"epoch": 1.1499381069442265,
"grad_norm": 0.2788243591785431,
"learning_rate": 5e-05,
"loss": 1.7748606204986572,
"step": 3298
},
{
"epoch": 1.150635493488153,
"grad_norm": 0.21569983661174774,
"learning_rate": 5e-05,
"loss": 1.9681100845336914,
"step": 3300
},
{
"epoch": 1.1513328800320797,
"grad_norm": 0.23890824615955353,
"learning_rate": 5e-05,
"loss": 1.9883267879486084,
"step": 3302
},
{
"epoch": 1.1520302665760065,
"grad_norm": 0.22588355839252472,
"learning_rate": 5e-05,
"loss": 1.9571938514709473,
"step": 3304
},
{
"epoch": 1.152727653119933,
"grad_norm": 0.3360740542411804,
"learning_rate": 5e-05,
"loss": 1.943509578704834,
"step": 3306
},
{
"epoch": 1.1534250396638597,
"grad_norm": 0.21990825235843658,
"learning_rate": 5e-05,
"loss": 1.8450465202331543,
"step": 3308
},
{
"epoch": 1.1541224262077863,
"grad_norm": 0.19391386210918427,
"learning_rate": 5e-05,
"loss": 1.7713053226470947,
"step": 3310
},
{
"epoch": 1.1548198127517129,
"grad_norm": 0.20962736010551453,
"learning_rate": 5e-05,
"loss": 1.9607583284378052,
"step": 3312
},
{
"epoch": 1.1555171992956397,
"grad_norm": 0.24098530411720276,
"learning_rate": 5e-05,
"loss": 2.0667452812194824,
"step": 3314
},
{
"epoch": 1.1562145858395663,
"grad_norm": 0.22169256210327148,
"learning_rate": 5e-05,
"loss": 1.8432475328445435,
"step": 3316
},
{
"epoch": 1.1569119723834929,
"grad_norm": 0.2243812382221222,
"learning_rate": 5e-05,
"loss": 1.9015934467315674,
"step": 3318
},
{
"epoch": 1.1576093589274195,
"grad_norm": 0.25815683603286743,
"learning_rate": 5e-05,
"loss": 1.8329825401306152,
"step": 3320
},
{
"epoch": 1.158306745471346,
"grad_norm": 0.23060393333435059,
"learning_rate": 5e-05,
"loss": 2.106444835662842,
"step": 3322
},
{
"epoch": 1.1590041320152729,
"grad_norm": 0.20663411915302277,
"learning_rate": 5e-05,
"loss": 1.9880495071411133,
"step": 3324
},
{
"epoch": 1.1597015185591995,
"grad_norm": 0.2329777330160141,
"learning_rate": 5e-05,
"loss": 2.0308775901794434,
"step": 3326
},
{
"epoch": 1.160398905103126,
"grad_norm": 0.1757218837738037,
"learning_rate": 5e-05,
"loss": 2.0551929473876953,
"step": 3328
},
{
"epoch": 1.1610962916470526,
"grad_norm": 0.23656487464904785,
"learning_rate": 4.995848899958489e-05,
"loss": 1.6389830112457275,
"step": 3330
},
{
"epoch": 1.1617936781909792,
"grad_norm": 0.23037122189998627,
"learning_rate": 4.991697799916978e-05,
"loss": 2.0219359397888184,
"step": 3332
},
{
"epoch": 1.162491064734906,
"grad_norm": 0.21388943493366241,
"learning_rate": 4.987546699875467e-05,
"loss": 1.992743730545044,
"step": 3334
},
{
"epoch": 1.1631884512788326,
"grad_norm": 0.1973319798707962,
"learning_rate": 4.983395599833956e-05,
"loss": 1.9138026237487793,
"step": 3336
},
{
"epoch": 1.1638858378227592,
"grad_norm": 0.20198167860507965,
"learning_rate": 4.9792444997924455e-05,
"loss": 1.9227473735809326,
"step": 3338
},
{
"epoch": 1.1645832243666858,
"grad_norm": 0.22222770750522614,
"learning_rate": 4.975093399750934e-05,
"loss": 2.050694465637207,
"step": 3340
},
{
"epoch": 1.1652806109106124,
"grad_norm": 0.18477162718772888,
"learning_rate": 4.9709422997094235e-05,
"loss": 1.860036849975586,
"step": 3342
},
{
"epoch": 1.1659779974545392,
"grad_norm": 0.19846336543560028,
"learning_rate": 4.966791199667912e-05,
"loss": 1.9285279512405396,
"step": 3344
},
{
"epoch": 1.1666753839984658,
"grad_norm": 0.23643462359905243,
"learning_rate": 4.962640099626401e-05,
"loss": 1.7353088855743408,
"step": 3346
},
{
"epoch": 1.1673727705423924,
"grad_norm": 0.24746716022491455,
"learning_rate": 4.95848899958489e-05,
"loss": 1.8223083019256592,
"step": 3348
},
{
"epoch": 1.168070157086319,
"grad_norm": 0.18461313843727112,
"learning_rate": 4.9543378995433794e-05,
"loss": 2.034353017807007,
"step": 3350
},
{
"epoch": 1.1687675436302456,
"grad_norm": 0.2313351333141327,
"learning_rate": 4.950186799501868e-05,
"loss": 1.6369261741638184,
"step": 3352
},
{
"epoch": 1.1694649301741724,
"grad_norm": 0.21167731285095215,
"learning_rate": 4.9460356994603574e-05,
"loss": 1.7105915546417236,
"step": 3354
},
{
"epoch": 1.170162316718099,
"grad_norm": 0.24277447164058685,
"learning_rate": 4.941884599418846e-05,
"loss": 1.6707457304000854,
"step": 3356
},
{
"epoch": 1.1708597032620256,
"grad_norm": 0.19461865723133087,
"learning_rate": 4.937733499377335e-05,
"loss": 1.8322867155075073,
"step": 3358
},
{
"epoch": 1.1715570898059522,
"grad_norm": 0.23283721506595612,
"learning_rate": 4.933582399335824e-05,
"loss": 1.699878454208374,
"step": 3360
},
{
"epoch": 1.1722544763498788,
"grad_norm": 0.2151813805103302,
"learning_rate": 4.9294312992943134e-05,
"loss": 1.6844890117645264,
"step": 3362
},
{
"epoch": 1.1729518628938054,
"grad_norm": 0.23241637647151947,
"learning_rate": 4.925280199252803e-05,
"loss": 1.8825969696044922,
"step": 3364
},
{
"epoch": 1.1736492494377322,
"grad_norm": 0.25263506174087524,
"learning_rate": 4.921129099211291e-05,
"loss": 1.5892612934112549,
"step": 3366
},
{
"epoch": 1.1743466359816588,
"grad_norm": 0.2107447236776352,
"learning_rate": 4.91697799916978e-05,
"loss": 2.078216552734375,
"step": 3368
},
{
"epoch": 1.1750440225255854,
"grad_norm": 0.17712299525737762,
"learning_rate": 4.912826899128269e-05,
"loss": 1.9671437740325928,
"step": 3370
},
{
"epoch": 1.175741409069512,
"grad_norm": 0.24711324274539948,
"learning_rate": 4.908675799086758e-05,
"loss": 1.8428316116333008,
"step": 3372
},
{
"epoch": 1.1764387956134386,
"grad_norm": 0.17892391979694366,
"learning_rate": 4.904524699045247e-05,
"loss": 2.0459518432617188,
"step": 3374
},
{
"epoch": 1.1771361821573652,
"grad_norm": 0.22361093759536743,
"learning_rate": 4.9003735990037366e-05,
"loss": 1.9006704092025757,
"step": 3376
},
{
"epoch": 1.177833568701292,
"grad_norm": 0.1812906414270401,
"learning_rate": 4.896222498962225e-05,
"loss": 2.1166930198669434,
"step": 3378
},
{
"epoch": 1.1785309552452186,
"grad_norm": 0.23143340647220612,
"learning_rate": 4.892071398920714e-05,
"loss": 2.0069873332977295,
"step": 3380
},
{
"epoch": 1.1792283417891452,
"grad_norm": 0.18367265164852142,
"learning_rate": 4.887920298879203e-05,
"loss": 1.997917890548706,
"step": 3382
},
{
"epoch": 1.1799257283330717,
"grad_norm": 0.20079566538333893,
"learning_rate": 4.883769198837692e-05,
"loss": 1.9160549640655518,
"step": 3384
},
{
"epoch": 1.1806231148769983,
"grad_norm": 0.21037310361862183,
"learning_rate": 4.879618098796181e-05,
"loss": 1.8643665313720703,
"step": 3386
},
{
"epoch": 1.1813205014209252,
"grad_norm": 0.1846771389245987,
"learning_rate": 4.8754669987546705e-05,
"loss": 1.8175299167633057,
"step": 3388
},
{
"epoch": 1.1820178879648517,
"grad_norm": 0.20625488460063934,
"learning_rate": 4.871315898713159e-05,
"loss": 1.7437139749526978,
"step": 3390
},
{
"epoch": 1.1827152745087783,
"grad_norm": 0.19118858873844147,
"learning_rate": 4.8671647986716485e-05,
"loss": 1.9128696918487549,
"step": 3392
},
{
"epoch": 1.183412661052705,
"grad_norm": 0.18890495598316193,
"learning_rate": 4.863013698630137e-05,
"loss": 2.0668766498565674,
"step": 3394
},
{
"epoch": 1.1841100475966315,
"grad_norm": 0.20588761568069458,
"learning_rate": 4.858862598588626e-05,
"loss": 1.9559204578399658,
"step": 3396
},
{
"epoch": 1.1848074341405583,
"grad_norm": 0.3991144895553589,
"learning_rate": 4.854711498547115e-05,
"loss": 1.866858720779419,
"step": 3398
},
{
"epoch": 1.185504820684485,
"grad_norm": 0.18666256964206696,
"learning_rate": 4.8505603985056045e-05,
"loss": 1.8842792510986328,
"step": 3400
},
{
"epoch": 1.1862022072284115,
"grad_norm": 0.18927165865898132,
"learning_rate": 4.846409298464093e-05,
"loss": 1.8684639930725098,
"step": 3402
},
{
"epoch": 1.1868995937723381,
"grad_norm": 0.3511774241924286,
"learning_rate": 4.8422581984225824e-05,
"loss": 1.7085282802581787,
"step": 3404
},
{
"epoch": 1.1875969803162647,
"grad_norm": 0.169032022356987,
"learning_rate": 4.838107098381071e-05,
"loss": 1.7434031963348389,
"step": 3406
},
{
"epoch": 1.1882943668601915,
"grad_norm": 0.18450136482715607,
"learning_rate": 4.8339559983395604e-05,
"loss": 1.7748292684555054,
"step": 3408
},
{
"epoch": 1.1889917534041181,
"grad_norm": 0.23342275619506836,
"learning_rate": 4.829804898298049e-05,
"loss": 2.0023927688598633,
"step": 3410
},
{
"epoch": 1.1896891399480447,
"grad_norm": 0.3139027953147888,
"learning_rate": 4.8256537982565384e-05,
"loss": 2.0264627933502197,
"step": 3412
},
{
"epoch": 1.1903865264919713,
"grad_norm": 0.2475530356168747,
"learning_rate": 4.821502698215028e-05,
"loss": 1.5474470853805542,
"step": 3414
},
{
"epoch": 1.191083913035898,
"grad_norm": 0.18853792548179626,
"learning_rate": 4.8173515981735164e-05,
"loss": 1.721402645111084,
"step": 3416
},
{
"epoch": 1.1917812995798247,
"grad_norm": 0.2692128121852875,
"learning_rate": 4.813200498132005e-05,
"loss": 1.8852906227111816,
"step": 3418
},
{
"epoch": 1.1924786861237513,
"grad_norm": 0.21304844319820404,
"learning_rate": 4.809049398090494e-05,
"loss": 1.7414357662200928,
"step": 3420
},
{
"epoch": 1.193176072667678,
"grad_norm": 0.19205497205257416,
"learning_rate": 4.804898298048983e-05,
"loss": 1.9370068311691284,
"step": 3422
},
{
"epoch": 1.1938734592116045,
"grad_norm": 0.2279967963695526,
"learning_rate": 4.800747198007472e-05,
"loss": 2.0227110385894775,
"step": 3424
},
{
"epoch": 1.194570845755531,
"grad_norm": 0.27181559801101685,
"learning_rate": 4.7965960979659616e-05,
"loss": 1.662685513496399,
"step": 3426
},
{
"epoch": 1.1952682322994579,
"grad_norm": 0.20728352665901184,
"learning_rate": 4.79244499792445e-05,
"loss": 1.686200499534607,
"step": 3428
},
{
"epoch": 1.1959656188433845,
"grad_norm": 0.2239687591791153,
"learning_rate": 4.7882938978829396e-05,
"loss": 1.9696838855743408,
"step": 3430
},
{
"epoch": 1.196663005387311,
"grad_norm": 0.18133199214935303,
"learning_rate": 4.784142797841428e-05,
"loss": 1.9330344200134277,
"step": 3432
},
{
"epoch": 1.1973603919312377,
"grad_norm": 0.2646746039390564,
"learning_rate": 4.779991697799917e-05,
"loss": 1.768394112586975,
"step": 3434
},
{
"epoch": 1.1980577784751643,
"grad_norm": 0.23626549541950226,
"learning_rate": 4.775840597758406e-05,
"loss": 1.870558500289917,
"step": 3436
},
{
"epoch": 1.198755165019091,
"grad_norm": 0.21471337974071503,
"learning_rate": 4.7716894977168955e-05,
"loss": 1.9804586172103882,
"step": 3438
},
{
"epoch": 1.1994525515630177,
"grad_norm": 0.1563311368227005,
"learning_rate": 4.767538397675384e-05,
"loss": 2.023649215698242,
"step": 3440
},
{
"epoch": 1.2001499381069443,
"grad_norm": 0.22683121263980865,
"learning_rate": 4.7633872976338735e-05,
"loss": 2.0708491802215576,
"step": 3442
},
{
"epoch": 1.2008473246508709,
"grad_norm": 0.20178870856761932,
"learning_rate": 4.759236197592362e-05,
"loss": 2.0271801948547363,
"step": 3444
},
{
"epoch": 1.2015447111947974,
"grad_norm": 0.2031419426202774,
"learning_rate": 4.755085097550851e-05,
"loss": 1.803966999053955,
"step": 3446
},
{
"epoch": 1.202242097738724,
"grad_norm": 0.20705781877040863,
"learning_rate": 4.75093399750934e-05,
"loss": 1.9386980533599854,
"step": 3448
},
{
"epoch": 1.2029394842826509,
"grad_norm": 0.23213481903076172,
"learning_rate": 4.746782897467829e-05,
"loss": 1.9492812156677246,
"step": 3450
},
{
"epoch": 1.2036368708265774,
"grad_norm": 0.32713228464126587,
"learning_rate": 4.742631797426319e-05,
"loss": 1.7684662342071533,
"step": 3452
},
{
"epoch": 1.204334257370504,
"grad_norm": 0.1834288090467453,
"learning_rate": 4.7384806973848074e-05,
"loss": 1.8972407579421997,
"step": 3454
},
{
"epoch": 1.2050316439144306,
"grad_norm": 0.16768479347229004,
"learning_rate": 4.734329597343296e-05,
"loss": 1.8771820068359375,
"step": 3456
},
{
"epoch": 1.2057290304583572,
"grad_norm": 0.24547205865383148,
"learning_rate": 4.7301784973017854e-05,
"loss": 1.8240361213684082,
"step": 3458
},
{
"epoch": 1.206426417002284,
"grad_norm": 0.19472764432430267,
"learning_rate": 4.726027397260274e-05,
"loss": 1.8432650566101074,
"step": 3460
},
{
"epoch": 1.2071238035462106,
"grad_norm": 0.22428664565086365,
"learning_rate": 4.721876297218763e-05,
"loss": 2.025120258331299,
"step": 3462
},
{
"epoch": 1.2078211900901372,
"grad_norm": 0.30077216029167175,
"learning_rate": 4.717725197177252e-05,
"loss": 1.9015809297561646,
"step": 3464
},
{
"epoch": 1.2085185766340638,
"grad_norm": 0.23884889483451843,
"learning_rate": 4.7135740971357414e-05,
"loss": 1.8485043048858643,
"step": 3466
},
{
"epoch": 1.2092159631779904,
"grad_norm": 0.25459370017051697,
"learning_rate": 4.70942299709423e-05,
"loss": 2.0250611305236816,
"step": 3468
},
{
"epoch": 1.209913349721917,
"grad_norm": 0.19823633134365082,
"learning_rate": 4.7052718970527193e-05,
"loss": 2.0509567260742188,
"step": 3470
},
{
"epoch": 1.2106107362658438,
"grad_norm": 0.18955479562282562,
"learning_rate": 4.701120797011208e-05,
"loss": 1.8645436763763428,
"step": 3472
},
{
"epoch": 1.2113081228097704,
"grad_norm": 0.21084818243980408,
"learning_rate": 4.696969696969697e-05,
"loss": 1.9099547863006592,
"step": 3474
},
{
"epoch": 1.212005509353697,
"grad_norm": 0.21796827018260956,
"learning_rate": 4.692818596928186e-05,
"loss": 1.7076760530471802,
"step": 3476
},
{
"epoch": 1.2127028958976236,
"grad_norm": 0.1779508888721466,
"learning_rate": 4.688667496886675e-05,
"loss": 1.7775758504867554,
"step": 3478
},
{
"epoch": 1.2134002824415502,
"grad_norm": 0.19905345141887665,
"learning_rate": 4.6845163968451646e-05,
"loss": 1.7629680633544922,
"step": 3480
},
{
"epoch": 1.214097668985477,
"grad_norm": 0.2048860490322113,
"learning_rate": 4.680365296803653e-05,
"loss": 2.185713291168213,
"step": 3482
},
{
"epoch": 1.2147950555294036,
"grad_norm": 0.19234156608581543,
"learning_rate": 4.676214196762142e-05,
"loss": 1.9074513912200928,
"step": 3484
},
{
"epoch": 1.2154924420733302,
"grad_norm": 0.248717799782753,
"learning_rate": 4.672063096720631e-05,
"loss": 1.9241571426391602,
"step": 3486
},
{
"epoch": 1.2161898286172568,
"grad_norm": 0.18128502368927002,
"learning_rate": 4.66791199667912e-05,
"loss": 2.0770812034606934,
"step": 3488
},
{
"epoch": 1.2168872151611834,
"grad_norm": 0.3450137972831726,
"learning_rate": 4.663760896637609e-05,
"loss": 1.824404239654541,
"step": 3490
},
{
"epoch": 1.2175846017051102,
"grad_norm": 0.2801811397075653,
"learning_rate": 4.6596097965960985e-05,
"loss": 1.7312917709350586,
"step": 3492
},
{
"epoch": 1.2182819882490368,
"grad_norm": 0.3181356191635132,
"learning_rate": 4.655458696554587e-05,
"loss": 1.796698808670044,
"step": 3494
},
{
"epoch": 1.2189793747929634,
"grad_norm": 0.24504268169403076,
"learning_rate": 4.6513075965130765e-05,
"loss": 2.0287394523620605,
"step": 3496
},
{
"epoch": 1.21967676133689,
"grad_norm": 0.2806799113750458,
"learning_rate": 4.647156496471565e-05,
"loss": 1.9487348794937134,
"step": 3498
},
{
"epoch": 1.2203741478808166,
"grad_norm": 0.21466997265815735,
"learning_rate": 4.643005396430054e-05,
"loss": 1.9784234762191772,
"step": 3500
},
{
"epoch": 1.2210715344247434,
"grad_norm": 0.23370575904846191,
"learning_rate": 4.638854296388543e-05,
"loss": 1.8271565437316895,
"step": 3502
},
{
"epoch": 1.22176892096867,
"grad_norm": 0.24749323725700378,
"learning_rate": 4.6347031963470325e-05,
"loss": 1.7034164667129517,
"step": 3504
},
{
"epoch": 1.2224663075125966,
"grad_norm": 0.23612868785858154,
"learning_rate": 4.630552096305521e-05,
"loss": 1.8277530670166016,
"step": 3506
},
{
"epoch": 1.2231636940565231,
"grad_norm": 0.20082780718803406,
"learning_rate": 4.6264009962640104e-05,
"loss": 1.9224843978881836,
"step": 3508
},
{
"epoch": 1.2238610806004497,
"grad_norm": 0.20833761990070343,
"learning_rate": 4.622249896222499e-05,
"loss": 1.8502240180969238,
"step": 3510
},
{
"epoch": 1.2245584671443766,
"grad_norm": 0.19105948507785797,
"learning_rate": 4.618098796180988e-05,
"loss": 1.8120299577713013,
"step": 3512
},
{
"epoch": 1.2252558536883031,
"grad_norm": 0.22227859497070312,
"learning_rate": 4.613947696139477e-05,
"loss": 1.936504602432251,
"step": 3514
},
{
"epoch": 1.2259532402322297,
"grad_norm": 0.29469016194343567,
"learning_rate": 4.6097965960979664e-05,
"loss": 1.8727209568023682,
"step": 3516
},
{
"epoch": 1.2266506267761563,
"grad_norm": 0.3146420121192932,
"learning_rate": 4.605645496056456e-05,
"loss": 1.6546387672424316,
"step": 3518
},
{
"epoch": 1.227348013320083,
"grad_norm": 0.18740369379520416,
"learning_rate": 4.6014943960149444e-05,
"loss": 2.0303993225097656,
"step": 3520
},
{
"epoch": 1.2280453998640097,
"grad_norm": 0.17977657914161682,
"learning_rate": 4.597343295973433e-05,
"loss": 1.9954206943511963,
"step": 3522
},
{
"epoch": 1.2287427864079363,
"grad_norm": 0.16073213517665863,
"learning_rate": 4.593192195931922e-05,
"loss": 1.941744089126587,
"step": 3524
},
{
"epoch": 1.229440172951863,
"grad_norm": 0.2085241675376892,
"learning_rate": 4.589041095890411e-05,
"loss": 1.7921476364135742,
"step": 3526
},
{
"epoch": 1.2301375594957895,
"grad_norm": 0.2387385368347168,
"learning_rate": 4.5848899958489e-05,
"loss": 1.6815814971923828,
"step": 3528
},
{
"epoch": 1.230834946039716,
"grad_norm": 0.17198093235492706,
"learning_rate": 4.5807388958073896e-05,
"loss": 1.906362533569336,
"step": 3530
},
{
"epoch": 1.231532332583643,
"grad_norm": 0.269122838973999,
"learning_rate": 4.576587795765878e-05,
"loss": 1.8721747398376465,
"step": 3532
},
{
"epoch": 1.2322297191275695,
"grad_norm": 0.2076372653245926,
"learning_rate": 4.572436695724367e-05,
"loss": 1.6538219451904297,
"step": 3534
},
{
"epoch": 1.232927105671496,
"grad_norm": 0.2273891717195511,
"learning_rate": 4.568285595682856e-05,
"loss": 2.1148486137390137,
"step": 3536
},
{
"epoch": 1.2336244922154227,
"grad_norm": 0.199555441737175,
"learning_rate": 4.564134495641345e-05,
"loss": 2.002974271774292,
"step": 3538
},
{
"epoch": 1.2343218787593493,
"grad_norm": 0.21863995492458344,
"learning_rate": 4.559983395599834e-05,
"loss": 1.804283857345581,
"step": 3540
},
{
"epoch": 1.2350192653032759,
"grad_norm": 0.22414664924144745,
"learning_rate": 4.5558322955583236e-05,
"loss": 1.97829008102417,
"step": 3542
},
{
"epoch": 1.2357166518472027,
"grad_norm": 0.204759418964386,
"learning_rate": 4.551681195516812e-05,
"loss": 1.9078149795532227,
"step": 3544
},
{
"epoch": 1.2364140383911293,
"grad_norm": 0.18208447098731995,
"learning_rate": 4.5475300954753015e-05,
"loss": 1.9248151779174805,
"step": 3546
},
{
"epoch": 1.2371114249350559,
"grad_norm": 0.18947017192840576,
"learning_rate": 4.54337899543379e-05,
"loss": 1.959487795829773,
"step": 3548
},
{
"epoch": 1.2378088114789825,
"grad_norm": 0.16036711633205414,
"learning_rate": 4.539227895392279e-05,
"loss": 2.02119517326355,
"step": 3550
},
{
"epoch": 1.238506198022909,
"grad_norm": 0.26236605644226074,
"learning_rate": 4.535076795350768e-05,
"loss": 2.0933468341827393,
"step": 3552
},
{
"epoch": 1.2392035845668357,
"grad_norm": 0.17580027878284454,
"learning_rate": 4.5309256953092575e-05,
"loss": 1.9626896381378174,
"step": 3554
},
{
"epoch": 1.2399009711107625,
"grad_norm": 0.2410041093826294,
"learning_rate": 4.526774595267746e-05,
"loss": 1.8230382204055786,
"step": 3556
},
{
"epoch": 1.240598357654689,
"grad_norm": 0.25834792852401733,
"learning_rate": 4.5226234952262355e-05,
"loss": 2.0001306533813477,
"step": 3558
},
{
"epoch": 1.2412957441986157,
"grad_norm": 0.1982775330543518,
"learning_rate": 4.518472395184724e-05,
"loss": 1.9554624557495117,
"step": 3560
},
{
"epoch": 1.2419931307425423,
"grad_norm": 0.2295491248369217,
"learning_rate": 4.514321295143213e-05,
"loss": 2.05601167678833,
"step": 3562
},
{
"epoch": 1.2426905172864688,
"grad_norm": 0.18504087626934052,
"learning_rate": 4.510170195101702e-05,
"loss": 1.8683438301086426,
"step": 3564
},
{
"epoch": 1.2433879038303957,
"grad_norm": 0.24018263816833496,
"learning_rate": 4.506019095060191e-05,
"loss": 1.8633451461791992,
"step": 3566
},
{
"epoch": 1.2440852903743223,
"grad_norm": 0.19507579505443573,
"learning_rate": 4.501867995018681e-05,
"loss": 2.019991874694824,
"step": 3568
},
{
"epoch": 1.2447826769182488,
"grad_norm": 0.19402845203876495,
"learning_rate": 4.4977168949771694e-05,
"loss": 1.9270744323730469,
"step": 3570
},
{
"epoch": 1.2454800634621754,
"grad_norm": 0.21988296508789062,
"learning_rate": 4.493565794935658e-05,
"loss": 1.9201555252075195,
"step": 3572
},
{
"epoch": 1.246177450006102,
"grad_norm": 0.16705384850502014,
"learning_rate": 4.4894146948941474e-05,
"loss": 1.8338549137115479,
"step": 3574
},
{
"epoch": 1.2468748365500288,
"grad_norm": 0.22458107769489288,
"learning_rate": 4.485263594852636e-05,
"loss": 1.6874667406082153,
"step": 3576
},
{
"epoch": 1.2475722230939554,
"grad_norm": 0.2693527340888977,
"learning_rate": 4.4811124948111246e-05,
"loss": 1.896040439605713,
"step": 3578
},
{
"epoch": 1.248269609637882,
"grad_norm": 0.2808539867401123,
"learning_rate": 4.476961394769614e-05,
"loss": 1.6780085563659668,
"step": 3580
},
{
"epoch": 1.2489669961818086,
"grad_norm": 0.1623006910085678,
"learning_rate": 4.472810294728103e-05,
"loss": 1.770935297012329,
"step": 3582
},
{
"epoch": 1.2496643827257352,
"grad_norm": 0.1795082837343216,
"learning_rate": 4.468659194686592e-05,
"loss": 1.901755690574646,
"step": 3584
},
{
"epoch": 1.250361769269662,
"grad_norm": 0.2932555377483368,
"learning_rate": 4.464508094645081e-05,
"loss": 1.9096486568450928,
"step": 3586
},
{
"epoch": 1.2510591558135886,
"grad_norm": 0.2195063978433609,
"learning_rate": 4.46035699460357e-05,
"loss": 2.0617709159851074,
"step": 3588
},
{
"epoch": 1.2517565423575152,
"grad_norm": 0.22299563884735107,
"learning_rate": 4.456205894562059e-05,
"loss": 2.0235157012939453,
"step": 3590
},
{
"epoch": 1.2524539289014418,
"grad_norm": 0.21385671198368073,
"learning_rate": 4.452054794520548e-05,
"loss": 1.802074909210205,
"step": 3592
},
{
"epoch": 1.2531513154453684,
"grad_norm": 0.19300207495689392,
"learning_rate": 4.447903694479037e-05,
"loss": 1.9281713962554932,
"step": 3594
},
{
"epoch": 1.2538487019892952,
"grad_norm": 0.20466646552085876,
"learning_rate": 4.4437525944375266e-05,
"loss": 1.8414555788040161,
"step": 3596
},
{
"epoch": 1.2545460885332218,
"grad_norm": 0.22405757009983063,
"learning_rate": 4.439601494396015e-05,
"loss": 1.7767560482025146,
"step": 3598
},
{
"epoch": 1.2552434750771484,
"grad_norm": 0.24353134632110596,
"learning_rate": 4.435450394354504e-05,
"loss": 1.7910246849060059,
"step": 3600
},
{
"epoch": 1.255940861621075,
"grad_norm": 0.2571335434913635,
"learning_rate": 4.431299294312993e-05,
"loss": 1.7598892450332642,
"step": 3602
},
{
"epoch": 1.2566382481650016,
"grad_norm": 0.17603053152561188,
"learning_rate": 4.427148194271482e-05,
"loss": 1.9008474349975586,
"step": 3604
},
{
"epoch": 1.2573356347089284,
"grad_norm": 0.239348903298378,
"learning_rate": 4.422997094229971e-05,
"loss": 2.0794053077697754,
"step": 3606
},
{
"epoch": 1.258033021252855,
"grad_norm": 0.1812250316143036,
"learning_rate": 4.4188459941884605e-05,
"loss": 2.1404871940612793,
"step": 3608
},
{
"epoch": 1.2587304077967816,
"grad_norm": 0.24487517774105072,
"learning_rate": 4.414694894146949e-05,
"loss": 1.8339680433273315,
"step": 3610
},
{
"epoch": 1.2594277943407082,
"grad_norm": 0.2815767228603363,
"learning_rate": 4.4105437941054384e-05,
"loss": 1.6045701503753662,
"step": 3612
},
{
"epoch": 1.2601251808846348,
"grad_norm": 0.22998550534248352,
"learning_rate": 4.406392694063927e-05,
"loss": 1.7449369430541992,
"step": 3614
},
{
"epoch": 1.2608225674285616,
"grad_norm": 0.20214541256427765,
"learning_rate": 4.402241594022416e-05,
"loss": 1.929940104484558,
"step": 3616
},
{
"epoch": 1.2615199539724882,
"grad_norm": 0.19958963990211487,
"learning_rate": 4.398090493980905e-05,
"loss": 2.0628929138183594,
"step": 3618
},
{
"epoch": 1.2622173405164148,
"grad_norm": 0.182153582572937,
"learning_rate": 4.3939393939393944e-05,
"loss": 1.8966150283813477,
"step": 3620
},
{
"epoch": 1.2629147270603414,
"grad_norm": 0.21103282272815704,
"learning_rate": 4.389788293897883e-05,
"loss": 2.0354502201080322,
"step": 3622
},
{
"epoch": 1.263612113604268,
"grad_norm": 0.20081013441085815,
"learning_rate": 4.3856371938563724e-05,
"loss": 1.7883808612823486,
"step": 3624
},
{
"epoch": 1.2643095001481948,
"grad_norm": 0.2175534963607788,
"learning_rate": 4.381486093814861e-05,
"loss": 1.7685527801513672,
"step": 3626
},
{
"epoch": 1.2650068866921214,
"grad_norm": 0.17973041534423828,
"learning_rate": 4.37733499377335e-05,
"loss": 2.0225534439086914,
"step": 3628
},
{
"epoch": 1.265704273236048,
"grad_norm": 0.15598313510417938,
"learning_rate": 4.373183893731839e-05,
"loss": 1.792747139930725,
"step": 3630
},
{
"epoch": 1.2664016597799745,
"grad_norm": 0.20121851563453674,
"learning_rate": 4.369032793690328e-05,
"loss": 1.8448128700256348,
"step": 3632
},
{
"epoch": 1.2670990463239011,
"grad_norm": 0.16615882515907288,
"learning_rate": 4.3648816936488176e-05,
"loss": 2.0251543521881104,
"step": 3634
},
{
"epoch": 1.267796432867828,
"grad_norm": 0.19637389481067657,
"learning_rate": 4.360730593607306e-05,
"loss": 2.016298294067383,
"step": 3636
},
{
"epoch": 1.2684938194117543,
"grad_norm": 0.1856715977191925,
"learning_rate": 4.356579493565795e-05,
"loss": 1.5390807390213013,
"step": 3638
},
{
"epoch": 1.2691912059556811,
"grad_norm": 0.2513745129108429,
"learning_rate": 4.352428393524284e-05,
"loss": 1.8810421228408813,
"step": 3640
},
{
"epoch": 1.2698885924996077,
"grad_norm": 0.18492180109024048,
"learning_rate": 4.348277293482773e-05,
"loss": 2.0095314979553223,
"step": 3642
},
{
"epoch": 1.2705859790435343,
"grad_norm": 0.18267939984798431,
"learning_rate": 4.344126193441262e-05,
"loss": 1.6570932865142822,
"step": 3644
},
{
"epoch": 1.271283365587461,
"grad_norm": 0.3277067542076111,
"learning_rate": 4.3399750933997516e-05,
"loss": 1.7094237804412842,
"step": 3646
},
{
"epoch": 1.2719807521313875,
"grad_norm": 0.26180464029312134,
"learning_rate": 4.33582399335824e-05,
"loss": 1.7283666133880615,
"step": 3648
},
{
"epoch": 1.2726781386753143,
"grad_norm": 0.30647602677345276,
"learning_rate": 4.331672893316729e-05,
"loss": 1.7029744386672974,
"step": 3650
},
{
"epoch": 1.273375525219241,
"grad_norm": 0.2590297758579254,
"learning_rate": 4.327521793275218e-05,
"loss": 2.0424137115478516,
"step": 3652
},
{
"epoch": 1.2740729117631675,
"grad_norm": 0.18071965873241425,
"learning_rate": 4.323370693233707e-05,
"loss": 1.8366672992706299,
"step": 3654
},
{
"epoch": 1.274770298307094,
"grad_norm": 0.2151574045419693,
"learning_rate": 4.319219593192196e-05,
"loss": 1.8460662364959717,
"step": 3656
},
{
"epoch": 1.2754676848510207,
"grad_norm": 0.21344274282455444,
"learning_rate": 4.3150684931506855e-05,
"loss": 1.8730554580688477,
"step": 3658
},
{
"epoch": 1.2761650713949475,
"grad_norm": 0.19345542788505554,
"learning_rate": 4.310917393109174e-05,
"loss": 1.9790599346160889,
"step": 3660
},
{
"epoch": 1.276862457938874,
"grad_norm": 0.2242473065853119,
"learning_rate": 4.3067662930676635e-05,
"loss": 1.7557787895202637,
"step": 3662
},
{
"epoch": 1.2775598444828007,
"grad_norm": 0.2564777135848999,
"learning_rate": 4.302615193026152e-05,
"loss": 2.153618335723877,
"step": 3664
},
{
"epoch": 1.2782572310267273,
"grad_norm": 0.1788969337940216,
"learning_rate": 4.298464092984641e-05,
"loss": 1.89314603805542,
"step": 3666
},
{
"epoch": 1.2789546175706539,
"grad_norm": 0.17710663378238678,
"learning_rate": 4.29431299294313e-05,
"loss": 1.7922167778015137,
"step": 3668
},
{
"epoch": 1.2796520041145807,
"grad_norm": 0.19768957793712616,
"learning_rate": 4.290161892901619e-05,
"loss": 1.9867340326309204,
"step": 3670
},
{
"epoch": 1.2803493906585073,
"grad_norm": 0.21366344392299652,
"learning_rate": 4.286010792860108e-05,
"loss": 1.7326983213424683,
"step": 3672
},
{
"epoch": 1.2810467772024339,
"grad_norm": 0.21440842747688293,
"learning_rate": 4.2818596928185974e-05,
"loss": 1.640990972518921,
"step": 3674
},
{
"epoch": 1.2817441637463605,
"grad_norm": 0.24721771478652954,
"learning_rate": 4.277708592777086e-05,
"loss": 1.8452787399291992,
"step": 3676
},
{
"epoch": 1.282441550290287,
"grad_norm": 0.19078348577022552,
"learning_rate": 4.2735574927355754e-05,
"loss": 1.9032864570617676,
"step": 3678
},
{
"epoch": 1.2831389368342139,
"grad_norm": 0.28183162212371826,
"learning_rate": 4.269406392694064e-05,
"loss": 1.7056008577346802,
"step": 3680
},
{
"epoch": 1.2838363233781405,
"grad_norm": 0.26828888058662415,
"learning_rate": 4.2652552926525527e-05,
"loss": 1.9642218351364136,
"step": 3682
},
{
"epoch": 1.284533709922067,
"grad_norm": 0.21766279637813568,
"learning_rate": 4.261104192611042e-05,
"loss": 1.9931907653808594,
"step": 3684
},
{
"epoch": 1.2852310964659936,
"grad_norm": 0.21355648338794708,
"learning_rate": 4.256953092569531e-05,
"loss": 1.7829322814941406,
"step": 3686
},
{
"epoch": 1.2859284830099202,
"grad_norm": 0.19305700063705444,
"learning_rate": 4.25280199252802e-05,
"loss": 1.9651380777359009,
"step": 3688
},
{
"epoch": 1.286625869553847,
"grad_norm": 0.1993425339460373,
"learning_rate": 4.248650892486509e-05,
"loss": 1.9956204891204834,
"step": 3690
},
{
"epoch": 1.2873232560977736,
"grad_norm": 0.20702120661735535,
"learning_rate": 4.244499792444998e-05,
"loss": 1.8387451171875,
"step": 3692
},
{
"epoch": 1.2880206426417002,
"grad_norm": 0.23407086730003357,
"learning_rate": 4.2403486924034866e-05,
"loss": 1.9918265342712402,
"step": 3694
},
{
"epoch": 1.2887180291856268,
"grad_norm": 0.26085415482521057,
"learning_rate": 4.236197592361976e-05,
"loss": 1.8425365686416626,
"step": 3696
},
{
"epoch": 1.2894154157295534,
"grad_norm": 0.21324673295021057,
"learning_rate": 4.232046492320465e-05,
"loss": 1.8807883262634277,
"step": 3698
},
{
"epoch": 1.2901128022734802,
"grad_norm": 0.19220253825187683,
"learning_rate": 4.2278953922789546e-05,
"loss": 2.061221122741699,
"step": 3700
},
{
"epoch": 1.2908101888174068,
"grad_norm": 0.24381768703460693,
"learning_rate": 4.223744292237443e-05,
"loss": 1.713374137878418,
"step": 3702
},
{
"epoch": 1.2915075753613334,
"grad_norm": 0.16915400326251984,
"learning_rate": 4.219593192195932e-05,
"loss": 2.04878306388855,
"step": 3704
},
{
"epoch": 1.29220496190526,
"grad_norm": 0.18365345895290375,
"learning_rate": 4.215442092154421e-05,
"loss": 1.8287684917449951,
"step": 3706
},
{
"epoch": 1.2929023484491866,
"grad_norm": 0.19457925856113434,
"learning_rate": 4.21129099211291e-05,
"loss": 1.7343847751617432,
"step": 3708
},
{
"epoch": 1.2935997349931134,
"grad_norm": 0.2683984339237213,
"learning_rate": 4.207139892071399e-05,
"loss": 1.885354995727539,
"step": 3710
},
{
"epoch": 1.29429712153704,
"grad_norm": 0.2496860921382904,
"learning_rate": 4.2029887920298885e-05,
"loss": 1.7821791172027588,
"step": 3712
},
{
"epoch": 1.2949945080809666,
"grad_norm": 0.19183088839054108,
"learning_rate": 4.198837691988377e-05,
"loss": 1.8185292482376099,
"step": 3714
},
{
"epoch": 1.2956918946248932,
"grad_norm": 0.23140807449817657,
"learning_rate": 4.194686591946866e-05,
"loss": 1.8875012397766113,
"step": 3716
},
{
"epoch": 1.2963892811688198,
"grad_norm": 0.20489057898521423,
"learning_rate": 4.190535491905355e-05,
"loss": 1.931618094444275,
"step": 3718
},
{
"epoch": 1.2970866677127466,
"grad_norm": 0.30073538422584534,
"learning_rate": 4.186384391863844e-05,
"loss": 1.6778407096862793,
"step": 3720
},
{
"epoch": 1.297784054256673,
"grad_norm": 0.20670728385448456,
"learning_rate": 4.182233291822333e-05,
"loss": 1.8695796728134155,
"step": 3722
},
{
"epoch": 1.2984814408005998,
"grad_norm": 0.18836669623851776,
"learning_rate": 4.1780821917808224e-05,
"loss": 1.9246810674667358,
"step": 3724
},
{
"epoch": 1.2991788273445264,
"grad_norm": 0.21102531254291534,
"learning_rate": 4.173931091739311e-05,
"loss": 1.5305893421173096,
"step": 3726
},
{
"epoch": 1.299876213888453,
"grad_norm": 0.22151389718055725,
"learning_rate": 4.1697799916978004e-05,
"loss": 1.6974706649780273,
"step": 3728
},
{
"epoch": 1.3005736004323798,
"grad_norm": 0.2221660166978836,
"learning_rate": 4.165628891656289e-05,
"loss": 1.6136248111724854,
"step": 3730
},
{
"epoch": 1.3012709869763062,
"grad_norm": 0.26578015089035034,
"learning_rate": 4.161477791614778e-05,
"loss": 1.9251880645751953,
"step": 3732
},
{
"epoch": 1.301968373520233,
"grad_norm": 0.22999724745750427,
"learning_rate": 4.157326691573267e-05,
"loss": 1.8556199073791504,
"step": 3734
},
{
"epoch": 1.3026657600641596,
"grad_norm": 0.23809760808944702,
"learning_rate": 4.153175591531756e-05,
"loss": 1.5907342433929443,
"step": 3736
},
{
"epoch": 1.3033631466080862,
"grad_norm": 0.1910601109266281,
"learning_rate": 4.149024491490245e-05,
"loss": 1.7277190685272217,
"step": 3738
},
{
"epoch": 1.3040605331520128,
"grad_norm": 0.22471824288368225,
"learning_rate": 4.144873391448734e-05,
"loss": 1.7609460353851318,
"step": 3740
},
{
"epoch": 1.3047579196959393,
"grad_norm": 0.15890686213970184,
"learning_rate": 4.140722291407223e-05,
"loss": 2.0207226276397705,
"step": 3742
},
{
"epoch": 1.3054553062398662,
"grad_norm": 0.2509796917438507,
"learning_rate": 4.136571191365712e-05,
"loss": 1.791619062423706,
"step": 3744
},
{
"epoch": 1.3061526927837928,
"grad_norm": 0.23045137524604797,
"learning_rate": 4.132420091324201e-05,
"loss": 1.8534510135650635,
"step": 3746
},
{
"epoch": 1.3068500793277193,
"grad_norm": 0.31268563866615295,
"learning_rate": 4.12826899128269e-05,
"loss": 1.848926305770874,
"step": 3748
},
{
"epoch": 1.307547465871646,
"grad_norm": 0.24398784339427948,
"learning_rate": 4.1241178912411796e-05,
"loss": 1.9457557201385498,
"step": 3750
},
{
"epoch": 1.3082448524155725,
"grad_norm": 0.226293683052063,
"learning_rate": 4.119966791199668e-05,
"loss": 1.9069085121154785,
"step": 3752
},
{
"epoch": 1.3089422389594993,
"grad_norm": 0.3214268088340759,
"learning_rate": 4.115815691158157e-05,
"loss": 1.693648099899292,
"step": 3754
},
{
"epoch": 1.309639625503426,
"grad_norm": 0.19223105907440186,
"learning_rate": 4.111664591116646e-05,
"loss": 1.971592903137207,
"step": 3756
},
{
"epoch": 1.3103370120473525,
"grad_norm": 0.22184304893016815,
"learning_rate": 4.107513491075135e-05,
"loss": 1.9897865056991577,
"step": 3758
},
{
"epoch": 1.3110343985912791,
"grad_norm": 0.2309533953666687,
"learning_rate": 4.103362391033624e-05,
"loss": 1.8181315660476685,
"step": 3760
},
{
"epoch": 1.3117317851352057,
"grad_norm": 0.24246546626091003,
"learning_rate": 4.0992112909921135e-05,
"loss": 1.964922547340393,
"step": 3762
},
{
"epoch": 1.3124291716791325,
"grad_norm": 0.23949795961380005,
"learning_rate": 4.095060190950602e-05,
"loss": 1.8070027828216553,
"step": 3764
},
{
"epoch": 1.3131265582230591,
"grad_norm": 0.2429887354373932,
"learning_rate": 4.0909090909090915e-05,
"loss": 1.6951407194137573,
"step": 3766
},
{
"epoch": 1.3138239447669857,
"grad_norm": 0.1676030457019806,
"learning_rate": 4.08675799086758e-05,
"loss": 1.9189460277557373,
"step": 3768
},
{
"epoch": 1.3145213313109123,
"grad_norm": 0.2941918969154358,
"learning_rate": 4.082606890826069e-05,
"loss": 1.9094812870025635,
"step": 3770
},
{
"epoch": 1.315218717854839,
"grad_norm": 0.23258011043071747,
"learning_rate": 4.078455790784558e-05,
"loss": 1.944237232208252,
"step": 3772
},
{
"epoch": 1.3159161043987657,
"grad_norm": 0.2276853621006012,
"learning_rate": 4.0743046907430474e-05,
"loss": 1.9636101722717285,
"step": 3774
},
{
"epoch": 1.3166134909426923,
"grad_norm": 0.19254475831985474,
"learning_rate": 4.070153590701536e-05,
"loss": 1.9854357242584229,
"step": 3776
},
{
"epoch": 1.317310877486619,
"grad_norm": 0.20987406373023987,
"learning_rate": 4.0660024906600254e-05,
"loss": 1.8651220798492432,
"step": 3778
},
{
"epoch": 1.3180082640305455,
"grad_norm": 0.21716980636119843,
"learning_rate": 4.061851390618514e-05,
"loss": 1.6945891380310059,
"step": 3780
},
{
"epoch": 1.318705650574472,
"grad_norm": 0.25951460003852844,
"learning_rate": 4.057700290577003e-05,
"loss": 1.835130214691162,
"step": 3782
},
{
"epoch": 1.319403037118399,
"grad_norm": 0.2451895922422409,
"learning_rate": 4.053549190535492e-05,
"loss": 1.8862169981002808,
"step": 3784
},
{
"epoch": 1.3201004236623255,
"grad_norm": 0.19390936195850372,
"learning_rate": 4.049398090493981e-05,
"loss": 1.852891206741333,
"step": 3786
},
{
"epoch": 1.320797810206252,
"grad_norm": 0.18305902183055878,
"learning_rate": 4.045246990452471e-05,
"loss": 1.7882800102233887,
"step": 3788
},
{
"epoch": 1.3214951967501787,
"grad_norm": 0.19043204188346863,
"learning_rate": 4.041095890410959e-05,
"loss": 1.9668065309524536,
"step": 3790
},
{
"epoch": 1.3221925832941053,
"grad_norm": 0.2826180160045624,
"learning_rate": 4.036944790369448e-05,
"loss": 1.761106014251709,
"step": 3792
},
{
"epoch": 1.322889969838032,
"grad_norm": 0.203342005610466,
"learning_rate": 4.032793690327937e-05,
"loss": 1.7443761825561523,
"step": 3794
},
{
"epoch": 1.3235873563819587,
"grad_norm": 0.2297106385231018,
"learning_rate": 4.028642590286426e-05,
"loss": 1.7752022743225098,
"step": 3796
},
{
"epoch": 1.3242847429258853,
"grad_norm": 0.25763577222824097,
"learning_rate": 4.0244914902449146e-05,
"loss": 1.8969886302947998,
"step": 3798
},
{
"epoch": 1.3249821294698119,
"grad_norm": 0.16434696316719055,
"learning_rate": 4.020340390203404e-05,
"loss": 2.026998519897461,
"step": 3800
},
{
"epoch": 1.3256795160137385,
"grad_norm": 0.23141427338123322,
"learning_rate": 4.016189290161893e-05,
"loss": 1.8267557621002197,
"step": 3802
},
{
"epoch": 1.3263769025576653,
"grad_norm": 0.19090408086776733,
"learning_rate": 4.012038190120382e-05,
"loss": 1.9565386772155762,
"step": 3804
},
{
"epoch": 1.3270742891015919,
"grad_norm": 0.16628874838352203,
"learning_rate": 4.007887090078871e-05,
"loss": 1.9619072675704956,
"step": 3806
},
{
"epoch": 1.3277716756455185,
"grad_norm": 0.2259003221988678,
"learning_rate": 4.00373599003736e-05,
"loss": 1.837080955505371,
"step": 3808
},
{
"epoch": 1.328469062189445,
"grad_norm": 0.22264313697814941,
"learning_rate": 3.999584889995849e-05,
"loss": 1.9822357892990112,
"step": 3810
},
{
"epoch": 1.3291664487333716,
"grad_norm": 0.23959025740623474,
"learning_rate": 3.995433789954338e-05,
"loss": 1.9450230598449707,
"step": 3812
},
{
"epoch": 1.3298638352772985,
"grad_norm": 0.28191155195236206,
"learning_rate": 3.991282689912827e-05,
"loss": 2.009995460510254,
"step": 3814
},
{
"epoch": 1.3305612218212248,
"grad_norm": 0.20822228491306305,
"learning_rate": 3.9871315898713165e-05,
"loss": 2.0478005409240723,
"step": 3816
},
{
"epoch": 1.3312586083651516,
"grad_norm": 0.2227475941181183,
"learning_rate": 3.982980489829805e-05,
"loss": 1.6045297384262085,
"step": 3818
},
{
"epoch": 1.3319559949090782,
"grad_norm": 0.19294404983520508,
"learning_rate": 3.978829389788294e-05,
"loss": 2.0358939170837402,
"step": 3820
},
{
"epoch": 1.3326533814530048,
"grad_norm": 0.17991949617862701,
"learning_rate": 3.974678289746783e-05,
"loss": 1.7805553674697876,
"step": 3822
},
{
"epoch": 1.3333507679969314,
"grad_norm": 0.17198731005191803,
"learning_rate": 3.970527189705272e-05,
"loss": 1.9175341129302979,
"step": 3824
},
{
"epoch": 1.334048154540858,
"grad_norm": 0.2148863524198532,
"learning_rate": 3.966376089663761e-05,
"loss": 1.8524353504180908,
"step": 3826
},
{
"epoch": 1.3347455410847848,
"grad_norm": 0.2367827296257019,
"learning_rate": 3.9622249896222504e-05,
"loss": 1.8672997951507568,
"step": 3828
},
{
"epoch": 1.3354429276287114,
"grad_norm": 0.2613336741924286,
"learning_rate": 3.958073889580739e-05,
"loss": 1.7679369449615479,
"step": 3830
},
{
"epoch": 1.336140314172638,
"grad_norm": 0.22326500713825226,
"learning_rate": 3.9539227895392284e-05,
"loss": 1.9833571910858154,
"step": 3832
},
{
"epoch": 1.3368377007165646,
"grad_norm": 0.19459055364131927,
"learning_rate": 3.949771689497717e-05,
"loss": 1.9010367393493652,
"step": 3834
},
{
"epoch": 1.3375350872604912,
"grad_norm": 0.21678787469863892,
"learning_rate": 3.945620589456206e-05,
"loss": 1.800331473350525,
"step": 3836
},
{
"epoch": 1.338232473804418,
"grad_norm": 0.23653283715248108,
"learning_rate": 3.941469489414695e-05,
"loss": 1.8341562747955322,
"step": 3838
},
{
"epoch": 1.3389298603483446,
"grad_norm": 0.29159465432167053,
"learning_rate": 3.937318389373184e-05,
"loss": 1.8640989065170288,
"step": 3840
},
{
"epoch": 1.3396272468922712,
"grad_norm": 0.3140033483505249,
"learning_rate": 3.933167289331673e-05,
"loss": 1.791996717453003,
"step": 3842
},
{
"epoch": 1.3403246334361978,
"grad_norm": 0.18970122933387756,
"learning_rate": 3.929016189290162e-05,
"loss": 1.9928334951400757,
"step": 3844
},
{
"epoch": 1.3410220199801244,
"grad_norm": 0.18842211365699768,
"learning_rate": 3.924865089248651e-05,
"loss": 1.8882253170013428,
"step": 3846
},
{
"epoch": 1.3417194065240512,
"grad_norm": 0.2726379930973053,
"learning_rate": 3.9207139892071396e-05,
"loss": 1.6094839572906494,
"step": 3848
},
{
"epoch": 1.3424167930679778,
"grad_norm": 0.2849552631378174,
"learning_rate": 3.916562889165629e-05,
"loss": 1.7770938873291016,
"step": 3850
},
{
"epoch": 1.3431141796119044,
"grad_norm": 0.16128021478652954,
"learning_rate": 3.912411789124118e-05,
"loss": 1.9565045833587646,
"step": 3852
},
{
"epoch": 1.343811566155831,
"grad_norm": 0.19331401586532593,
"learning_rate": 3.9082606890826076e-05,
"loss": 1.9027246236801147,
"step": 3854
},
{
"epoch": 1.3445089526997576,
"grad_norm": 0.3598761558532715,
"learning_rate": 3.904109589041096e-05,
"loss": 1.4475805759429932,
"step": 3856
},
{
"epoch": 1.3452063392436844,
"grad_norm": 0.28386190533638,
"learning_rate": 3.899958488999585e-05,
"loss": 1.8573615550994873,
"step": 3858
},
{
"epoch": 1.345903725787611,
"grad_norm": 0.28555598855018616,
"learning_rate": 3.895807388958074e-05,
"loss": 1.7625099420547485,
"step": 3860
},
{
"epoch": 1.3466011123315376,
"grad_norm": 0.2793368697166443,
"learning_rate": 3.891656288916563e-05,
"loss": 1.790490746498108,
"step": 3862
},
{
"epoch": 1.3472984988754642,
"grad_norm": 0.18027624487876892,
"learning_rate": 3.887505188875052e-05,
"loss": 2.063419818878174,
"step": 3864
},
{
"epoch": 1.3479958854193907,
"grad_norm": 0.18590649962425232,
"learning_rate": 3.8833540888335415e-05,
"loss": 1.8247560262680054,
"step": 3866
},
{
"epoch": 1.3486932719633176,
"grad_norm": 0.15488718450069427,
"learning_rate": 3.87920298879203e-05,
"loss": 2.0603365898132324,
"step": 3868
},
{
"epoch": 1.3493906585072442,
"grad_norm": 0.2252766340970993,
"learning_rate": 3.875051888750519e-05,
"loss": 2.0061049461364746,
"step": 3870
},
{
"epoch": 1.3500880450511707,
"grad_norm": 0.18663561344146729,
"learning_rate": 3.870900788709008e-05,
"loss": 2.091830253601074,
"step": 3872
},
{
"epoch": 1.3507854315950973,
"grad_norm": 0.25875183939933777,
"learning_rate": 3.866749688667497e-05,
"loss": 2.0235438346862793,
"step": 3874
},
{
"epoch": 1.351482818139024,
"grad_norm": 0.169276624917984,
"learning_rate": 3.862598588625986e-05,
"loss": 1.8578166961669922,
"step": 3876
},
{
"epoch": 1.3521802046829507,
"grad_norm": 0.24092863500118256,
"learning_rate": 3.8584474885844754e-05,
"loss": 1.908102035522461,
"step": 3878
},
{
"epoch": 1.3528775912268773,
"grad_norm": 0.27469509840011597,
"learning_rate": 3.854296388542964e-05,
"loss": 2.037529468536377,
"step": 3880
},
{
"epoch": 1.353574977770804,
"grad_norm": 0.3646101951599121,
"learning_rate": 3.8501452885014534e-05,
"loss": 1.8467917442321777,
"step": 3882
},
{
"epoch": 1.3542723643147305,
"grad_norm": 0.24833038449287415,
"learning_rate": 3.845994188459942e-05,
"loss": 2.0194578170776367,
"step": 3884
},
{
"epoch": 1.3549697508586571,
"grad_norm": 0.17717555165290833,
"learning_rate": 3.841843088418431e-05,
"loss": 2.025177478790283,
"step": 3886
},
{
"epoch": 1.355667137402584,
"grad_norm": 0.18158070743083954,
"learning_rate": 3.83769198837692e-05,
"loss": 1.8164186477661133,
"step": 3888
},
{
"epoch": 1.3563645239465105,
"grad_norm": 0.34645575284957886,
"learning_rate": 3.8335408883354094e-05,
"loss": 1.7632312774658203,
"step": 3890
},
{
"epoch": 1.3570619104904371,
"grad_norm": 0.23498892784118652,
"learning_rate": 3.829389788293898e-05,
"loss": 1.6641490459442139,
"step": 3892
},
{
"epoch": 1.3577592970343637,
"grad_norm": 0.25509336590766907,
"learning_rate": 3.825238688252387e-05,
"loss": 1.8427915573120117,
"step": 3894
},
{
"epoch": 1.3584566835782903,
"grad_norm": 0.24074427783489227,
"learning_rate": 3.821087588210876e-05,
"loss": 1.8809864521026611,
"step": 3896
},
{
"epoch": 1.3591540701222171,
"grad_norm": 0.2611803710460663,
"learning_rate": 3.816936488169365e-05,
"loss": 1.7561538219451904,
"step": 3898
},
{
"epoch": 1.3598514566661437,
"grad_norm": 0.21703186631202698,
"learning_rate": 3.812785388127854e-05,
"loss": 1.8782120943069458,
"step": 3900
},
{
"epoch": 1.3605488432100703,
"grad_norm": 0.16963045299053192,
"learning_rate": 3.8086342880863426e-05,
"loss": 2.000520706176758,
"step": 3902
},
{
"epoch": 1.361246229753997,
"grad_norm": 0.19722816348075867,
"learning_rate": 3.8044831880448326e-05,
"loss": 1.783744215965271,
"step": 3904
},
{
"epoch": 1.3619436162979235,
"grad_norm": 0.19367583096027374,
"learning_rate": 3.800332088003321e-05,
"loss": 1.9716098308563232,
"step": 3906
},
{
"epoch": 1.3626410028418503,
"grad_norm": 0.2042233645915985,
"learning_rate": 3.79618098796181e-05,
"loss": 1.847687005996704,
"step": 3908
},
{
"epoch": 1.3633383893857767,
"grad_norm": 0.19127587974071503,
"learning_rate": 3.792029887920299e-05,
"loss": 2.027930736541748,
"step": 3910
},
{
"epoch": 1.3640357759297035,
"grad_norm": 0.23919139802455902,
"learning_rate": 3.787878787878788e-05,
"loss": 1.8762127161026,
"step": 3912
},
{
"epoch": 1.36473316247363,
"grad_norm": 0.21982508897781372,
"learning_rate": 3.7837276878372765e-05,
"loss": 2.0291380882263184,
"step": 3914
},
{
"epoch": 1.3654305490175567,
"grad_norm": 0.16975975036621094,
"learning_rate": 3.779576587795766e-05,
"loss": 1.933445692062378,
"step": 3916
},
{
"epoch": 1.3661279355614833,
"grad_norm": 0.21642948687076569,
"learning_rate": 3.775425487754255e-05,
"loss": 2.02642822265625,
"step": 3918
},
{
"epoch": 1.3668253221054099,
"grad_norm": 0.1782580465078354,
"learning_rate": 3.7712743877127445e-05,
"loss": 1.8347392082214355,
"step": 3920
},
{
"epoch": 1.3675227086493367,
"grad_norm": 0.19644644856452942,
"learning_rate": 3.767123287671233e-05,
"loss": 1.7302055358886719,
"step": 3922
},
{
"epoch": 1.3682200951932633,
"grad_norm": 0.16771207749843597,
"learning_rate": 3.762972187629722e-05,
"loss": 1.8333760499954224,
"step": 3924
},
{
"epoch": 1.3689174817371899,
"grad_norm": 0.2550005614757538,
"learning_rate": 3.758821087588211e-05,
"loss": 1.7815848588943481,
"step": 3926
},
{
"epoch": 1.3696148682811164,
"grad_norm": 0.24076442420482635,
"learning_rate": 3.7546699875467e-05,
"loss": 2.0017809867858887,
"step": 3928
},
{
"epoch": 1.370312254825043,
"grad_norm": 0.19728711247444153,
"learning_rate": 3.750518887505189e-05,
"loss": 1.8977465629577637,
"step": 3930
},
{
"epoch": 1.3710096413689699,
"grad_norm": 0.2591626048088074,
"learning_rate": 3.7463677874636784e-05,
"loss": 1.7393373250961304,
"step": 3932
},
{
"epoch": 1.3717070279128964,
"grad_norm": 0.1811761111021042,
"learning_rate": 3.742216687422167e-05,
"loss": 2.018988609313965,
"step": 3934
},
{
"epoch": 1.372404414456823,
"grad_norm": 0.22588220238685608,
"learning_rate": 3.738065587380656e-05,
"loss": 1.8173197507858276,
"step": 3936
},
{
"epoch": 1.3731018010007496,
"grad_norm": 0.19720309972763062,
"learning_rate": 3.733914487339145e-05,
"loss": 1.7693676948547363,
"step": 3938
},
{
"epoch": 1.3737991875446762,
"grad_norm": 0.27839139103889465,
"learning_rate": 3.729763387297634e-05,
"loss": 1.7554996013641357,
"step": 3940
},
{
"epoch": 1.374496574088603,
"grad_norm": 0.18104492127895355,
"learning_rate": 3.725612287256123e-05,
"loss": 2.034649133682251,
"step": 3942
},
{
"epoch": 1.3751939606325296,
"grad_norm": 0.21200129389762878,
"learning_rate": 3.7214611872146123e-05,
"loss": 1.7810739278793335,
"step": 3944
},
{
"epoch": 1.3758913471764562,
"grad_norm": 0.16660650074481964,
"learning_rate": 3.717310087173101e-05,
"loss": 2.110220432281494,
"step": 3946
},
{
"epoch": 1.3765887337203828,
"grad_norm": 0.24682077765464783,
"learning_rate": 3.71315898713159e-05,
"loss": 1.8052815198898315,
"step": 3948
},
{
"epoch": 1.3772861202643094,
"grad_norm": 0.18706710636615753,
"learning_rate": 3.709007887090079e-05,
"loss": 1.9499850273132324,
"step": 3950
},
{
"epoch": 1.3779835068082362,
"grad_norm": 0.2659950852394104,
"learning_rate": 3.7048567870485676e-05,
"loss": 1.854794979095459,
"step": 3952
},
{
"epoch": 1.3786808933521628,
"grad_norm": 0.20848090946674347,
"learning_rate": 3.700705687007057e-05,
"loss": 2.0630273818969727,
"step": 3954
},
{
"epoch": 1.3793782798960894,
"grad_norm": 0.26349514722824097,
"learning_rate": 3.696554586965546e-05,
"loss": 1.922311782836914,
"step": 3956
},
{
"epoch": 1.380075666440016,
"grad_norm": 0.2556496560573578,
"learning_rate": 3.692403486924035e-05,
"loss": 1.896592140197754,
"step": 3958
},
{
"epoch": 1.3807730529839426,
"grad_norm": 0.2060481160879135,
"learning_rate": 3.688252386882524e-05,
"loss": 1.9381667375564575,
"step": 3960
},
{
"epoch": 1.3814704395278694,
"grad_norm": 0.2973029911518097,
"learning_rate": 3.684101286841013e-05,
"loss": 1.9297459125518799,
"step": 3962
},
{
"epoch": 1.382167826071796,
"grad_norm": 0.20299437642097473,
"learning_rate": 3.679950186799502e-05,
"loss": 1.8177218437194824,
"step": 3964
},
{
"epoch": 1.3828652126157226,
"grad_norm": 0.2874215245246887,
"learning_rate": 3.675799086757991e-05,
"loss": 1.7827160358428955,
"step": 3966
},
{
"epoch": 1.3835625991596492,
"grad_norm": 0.17761044204235077,
"learning_rate": 3.67164798671648e-05,
"loss": 1.9967944622039795,
"step": 3968
},
{
"epoch": 1.3842599857035758,
"grad_norm": 0.19483721256256104,
"learning_rate": 3.6674968866749695e-05,
"loss": 1.7855472564697266,
"step": 3970
},
{
"epoch": 1.3849573722475026,
"grad_norm": 0.2172488272190094,
"learning_rate": 3.663345786633458e-05,
"loss": 1.9099082946777344,
"step": 3972
},
{
"epoch": 1.3856547587914292,
"grad_norm": 0.18784500658512115,
"learning_rate": 3.659194686591947e-05,
"loss": 2.0319933891296387,
"step": 3974
},
{
"epoch": 1.3863521453353558,
"grad_norm": 0.1919984221458435,
"learning_rate": 3.655043586550436e-05,
"loss": 1.8924497365951538,
"step": 3976
},
{
"epoch": 1.3870495318792824,
"grad_norm": 0.17610539495944977,
"learning_rate": 3.650892486508925e-05,
"loss": 1.715224266052246,
"step": 3978
},
{
"epoch": 1.387746918423209,
"grad_norm": 0.2167578637599945,
"learning_rate": 3.646741386467414e-05,
"loss": 2.06508731842041,
"step": 3980
},
{
"epoch": 1.3884443049671358,
"grad_norm": 0.16061186790466309,
"learning_rate": 3.6425902864259034e-05,
"loss": 1.7508708238601685,
"step": 3982
},
{
"epoch": 1.3891416915110624,
"grad_norm": 0.4638068377971649,
"learning_rate": 3.638439186384392e-05,
"loss": 1.4626554250717163,
"step": 3984
},
{
"epoch": 1.389839078054989,
"grad_norm": 0.22213895618915558,
"learning_rate": 3.634288086342881e-05,
"loss": 1.9912350177764893,
"step": 3986
},
{
"epoch": 1.3905364645989156,
"grad_norm": 0.2649343013763428,
"learning_rate": 3.63013698630137e-05,
"loss": 1.769735336303711,
"step": 3988
},
{
"epoch": 1.3912338511428421,
"grad_norm": 0.22359298169612885,
"learning_rate": 3.625985886259859e-05,
"loss": 2.0618157386779785,
"step": 3990
},
{
"epoch": 1.391931237686769,
"grad_norm": 0.20694513618946075,
"learning_rate": 3.621834786218348e-05,
"loss": 1.9176592826843262,
"step": 3992
},
{
"epoch": 1.3926286242306953,
"grad_norm": 0.19296231865882874,
"learning_rate": 3.6176836861768374e-05,
"loss": 1.7908153533935547,
"step": 3994
},
{
"epoch": 1.3933260107746221,
"grad_norm": 0.34000447392463684,
"learning_rate": 3.613532586135326e-05,
"loss": 1.942549467086792,
"step": 3996
},
{
"epoch": 1.3940233973185487,
"grad_norm": 0.15586970746517181,
"learning_rate": 3.6093814860938153e-05,
"loss": 1.8800139427185059,
"step": 3998
},
{
"epoch": 1.3947207838624753,
"grad_norm": 0.2056552916765213,
"learning_rate": 3.605230386052304e-05,
"loss": 1.8940651416778564,
"step": 4000
},
{
"epoch": 1.395418170406402,
"grad_norm": 0.33208680152893066,
"learning_rate": 3.6010792860107926e-05,
"loss": 1.785973072052002,
"step": 4002
},
{
"epoch": 1.3961155569503285,
"grad_norm": 0.27164021134376526,
"learning_rate": 3.596928185969282e-05,
"loss": 1.8067493438720703,
"step": 4004
},
{
"epoch": 1.3968129434942553,
"grad_norm": 0.21248085796833038,
"learning_rate": 3.592777085927771e-05,
"loss": 1.8170571327209473,
"step": 4006
},
{
"epoch": 1.397510330038182,
"grad_norm": 0.24447138607501984,
"learning_rate": 3.58862598588626e-05,
"loss": 1.8484935760498047,
"step": 4008
},
{
"epoch": 1.3982077165821085,
"grad_norm": 0.2377743124961853,
"learning_rate": 3.584474885844749e-05,
"loss": 2.105147123336792,
"step": 4010
},
{
"epoch": 1.398905103126035,
"grad_norm": 0.19377318024635315,
"learning_rate": 3.580323785803238e-05,
"loss": 1.8844549655914307,
"step": 4012
},
{
"epoch": 1.3996024896699617,
"grad_norm": 0.1808302253484726,
"learning_rate": 3.576172685761727e-05,
"loss": 1.7806236743927002,
"step": 4014
},
{
"epoch": 1.4002998762138885,
"grad_norm": 0.22681190073490143,
"learning_rate": 3.572021585720216e-05,
"loss": 1.9029309749603271,
"step": 4016
},
{
"epoch": 1.400997262757815,
"grad_norm": 0.24358852207660675,
"learning_rate": 3.5678704856787045e-05,
"loss": 2.170813798904419,
"step": 4018
},
{
"epoch": 1.4016946493017417,
"grad_norm": 0.16216611862182617,
"learning_rate": 3.5637193856371945e-05,
"loss": 1.963215708732605,
"step": 4020
},
{
"epoch": 1.4023920358456683,
"grad_norm": 0.1840636134147644,
"learning_rate": 3.559568285595683e-05,
"loss": 1.8479241132736206,
"step": 4022
},
{
"epoch": 1.4030894223895949,
"grad_norm": 0.20008444786071777,
"learning_rate": 3.555417185554172e-05,
"loss": 1.6921862363815308,
"step": 4024
},
{
"epoch": 1.4037868089335217,
"grad_norm": 0.19347982108592987,
"learning_rate": 3.551266085512661e-05,
"loss": 1.75076425075531,
"step": 4026
},
{
"epoch": 1.4044841954774483,
"grad_norm": 0.22057102620601654,
"learning_rate": 3.54711498547115e-05,
"loss": 1.9814503192901611,
"step": 4028
},
{
"epoch": 1.4051815820213749,
"grad_norm": 0.20400013029575348,
"learning_rate": 3.5429638854296385e-05,
"loss": 1.83992600440979,
"step": 4030
},
{
"epoch": 1.4058789685653015,
"grad_norm": 0.21387846767902374,
"learning_rate": 3.538812785388128e-05,
"loss": 1.9891995191574097,
"step": 4032
},
{
"epoch": 1.406576355109228,
"grad_norm": 0.2546032965183258,
"learning_rate": 3.534661685346617e-05,
"loss": 1.8944637775421143,
"step": 4034
},
{
"epoch": 1.4072737416531549,
"grad_norm": 0.214521586894989,
"learning_rate": 3.5305105853051064e-05,
"loss": 1.8203704357147217,
"step": 4036
},
{
"epoch": 1.4079711281970815,
"grad_norm": 0.21132892370224,
"learning_rate": 3.526359485263595e-05,
"loss": 1.85984468460083,
"step": 4038
},
{
"epoch": 1.408668514741008,
"grad_norm": 0.20359593629837036,
"learning_rate": 3.522208385222084e-05,
"loss": 1.8268412351608276,
"step": 4040
},
{
"epoch": 1.4093659012849347,
"grad_norm": 0.21819265186786652,
"learning_rate": 3.518057285180573e-05,
"loss": 1.7824187278747559,
"step": 4042
},
{
"epoch": 1.4100632878288613,
"grad_norm": 0.17693446576595306,
"learning_rate": 3.513906185139062e-05,
"loss": 1.7772114276885986,
"step": 4044
},
{
"epoch": 1.410760674372788,
"grad_norm": 0.26175910234451294,
"learning_rate": 3.509755085097551e-05,
"loss": 2.01741361618042,
"step": 4046
},
{
"epoch": 1.4114580609167147,
"grad_norm": 0.4403669834136963,
"learning_rate": 3.5056039850560404e-05,
"loss": 1.5121135711669922,
"step": 4048
},
{
"epoch": 1.4121554474606413,
"grad_norm": 0.1782636046409607,
"learning_rate": 3.501452885014529e-05,
"loss": 2.056703567504883,
"step": 4050
},
{
"epoch": 1.4128528340045678,
"grad_norm": 0.22794538736343384,
"learning_rate": 3.4973017849730177e-05,
"loss": 1.6770656108856201,
"step": 4052
},
{
"epoch": 1.4135502205484944,
"grad_norm": 0.2163773626089096,
"learning_rate": 3.493150684931507e-05,
"loss": 1.9402332305908203,
"step": 4054
},
{
"epoch": 1.4142476070924213,
"grad_norm": 0.23980838060379028,
"learning_rate": 3.4889995848899956e-05,
"loss": 2.0581789016723633,
"step": 4056
},
{
"epoch": 1.4149449936363478,
"grad_norm": 0.16321487724781036,
"learning_rate": 3.484848484848485e-05,
"loss": 2.007432222366333,
"step": 4058
},
{
"epoch": 1.4156423801802744,
"grad_norm": 0.22912773489952087,
"learning_rate": 3.480697384806974e-05,
"loss": 1.9420256614685059,
"step": 4060
},
{
"epoch": 1.416339766724201,
"grad_norm": 0.18415477871894836,
"learning_rate": 3.476546284765463e-05,
"loss": 1.902016282081604,
"step": 4062
},
{
"epoch": 1.4170371532681276,
"grad_norm": 0.2103758454322815,
"learning_rate": 3.472395184723952e-05,
"loss": 2.004079818725586,
"step": 4064
},
{
"epoch": 1.4177345398120544,
"grad_norm": 0.19446878135204315,
"learning_rate": 3.468244084682441e-05,
"loss": 1.9384950399398804,
"step": 4066
},
{
"epoch": 1.418431926355981,
"grad_norm": 0.22194212675094604,
"learning_rate": 3.4640929846409296e-05,
"loss": 1.9696013927459717,
"step": 4068
},
{
"epoch": 1.4191293128999076,
"grad_norm": 0.1842203140258789,
"learning_rate": 3.459941884599419e-05,
"loss": 1.9772635698318481,
"step": 4070
},
{
"epoch": 1.4198266994438342,
"grad_norm": 0.23830775916576385,
"learning_rate": 3.455790784557908e-05,
"loss": 1.716020107269287,
"step": 4072
},
{
"epoch": 1.4205240859877608,
"grad_norm": 0.3197784125804901,
"learning_rate": 3.451639684516397e-05,
"loss": 2.0234553813934326,
"step": 4074
},
{
"epoch": 1.4212214725316876,
"grad_norm": 0.24736444652080536,
"learning_rate": 3.447488584474886e-05,
"loss": 2.014583110809326,
"step": 4076
},
{
"epoch": 1.4219188590756142,
"grad_norm": 0.18203675746917725,
"learning_rate": 3.443337484433375e-05,
"loss": 2.062819480895996,
"step": 4078
},
{
"epoch": 1.4226162456195408,
"grad_norm": 0.34547385573387146,
"learning_rate": 3.439186384391864e-05,
"loss": 1.8202629089355469,
"step": 4080
},
{
"epoch": 1.4233136321634674,
"grad_norm": 0.2565050721168518,
"learning_rate": 3.435035284350353e-05,
"loss": 1.916255235671997,
"step": 4082
},
{
"epoch": 1.424011018707394,
"grad_norm": 0.2017725557088852,
"learning_rate": 3.430884184308842e-05,
"loss": 1.856203556060791,
"step": 4084
},
{
"epoch": 1.4247084052513208,
"grad_norm": 0.17718103528022766,
"learning_rate": 3.4267330842673315e-05,
"loss": 1.9452753067016602,
"step": 4086
},
{
"epoch": 1.4254057917952472,
"grad_norm": 0.22691674530506134,
"learning_rate": 3.42258198422582e-05,
"loss": 1.9916186332702637,
"step": 4088
},
{
"epoch": 1.426103178339174,
"grad_norm": 0.2360876351594925,
"learning_rate": 3.418430884184309e-05,
"loss": 1.9206513166427612,
"step": 4090
},
{
"epoch": 1.4268005648831006,
"grad_norm": 0.21341641247272491,
"learning_rate": 3.414279784142798e-05,
"loss": 1.9444701671600342,
"step": 4092
},
{
"epoch": 1.4274979514270272,
"grad_norm": 0.20100471377372742,
"learning_rate": 3.410128684101287e-05,
"loss": 1.8804447650909424,
"step": 4094
},
{
"epoch": 1.4281953379709538,
"grad_norm": 0.2367202192544937,
"learning_rate": 3.405977584059776e-05,
"loss": 1.7431578636169434,
"step": 4096
},
{
"epoch": 1.4288927245148804,
"grad_norm": 0.17000959813594818,
"learning_rate": 3.4018264840182654e-05,
"loss": 1.6879067420959473,
"step": 4098
},
{
"epoch": 1.4295901110588072,
"grad_norm": 0.23608246445655823,
"learning_rate": 3.397675383976754e-05,
"loss": 1.7276872396469116,
"step": 4100
},
{
"epoch": 1.4302874976027338,
"grad_norm": 0.1636294573545456,
"learning_rate": 3.3935242839352433e-05,
"loss": 1.8374544382095337,
"step": 4102
},
{
"epoch": 1.4309848841466604,
"grad_norm": 0.27517223358154297,
"learning_rate": 3.389373183893732e-05,
"loss": 1.8124754428863525,
"step": 4104
},
{
"epoch": 1.431682270690587,
"grad_norm": 0.17434199154376984,
"learning_rate": 3.3852220838522206e-05,
"loss": 1.9215025901794434,
"step": 4106
},
{
"epoch": 1.4323796572345135,
"grad_norm": 0.2249208241701126,
"learning_rate": 3.38107098381071e-05,
"loss": 1.9626878499984741,
"step": 4108
},
{
"epoch": 1.4330770437784404,
"grad_norm": 0.22073271870613098,
"learning_rate": 3.376919883769199e-05,
"loss": 1.8417041301727295,
"step": 4110
},
{
"epoch": 1.433774430322367,
"grad_norm": 0.231496199965477,
"learning_rate": 3.372768783727688e-05,
"loss": 1.8525700569152832,
"step": 4112
},
{
"epoch": 1.4344718168662935,
"grad_norm": 0.20072703063488007,
"learning_rate": 3.368617683686177e-05,
"loss": 1.935911774635315,
"step": 4114
},
{
"epoch": 1.4351692034102201,
"grad_norm": 0.19549134373664856,
"learning_rate": 3.364466583644666e-05,
"loss": 1.8967294692993164,
"step": 4116
},
{
"epoch": 1.4358665899541467,
"grad_norm": 0.2024395614862442,
"learning_rate": 3.3603154836031546e-05,
"loss": 1.7203136682510376,
"step": 4118
},
{
"epoch": 1.4365639764980735,
"grad_norm": 0.16281065344810486,
"learning_rate": 3.356164383561644e-05,
"loss": 1.841869831085205,
"step": 4120
},
{
"epoch": 1.4372613630420001,
"grad_norm": 0.20409001410007477,
"learning_rate": 3.352013283520133e-05,
"loss": 2.0007100105285645,
"step": 4122
},
{
"epoch": 1.4379587495859267,
"grad_norm": 0.18246321380138397,
"learning_rate": 3.3478621834786225e-05,
"loss": 1.8394222259521484,
"step": 4124
},
{
"epoch": 1.4386561361298533,
"grad_norm": 0.17156104743480682,
"learning_rate": 3.343711083437111e-05,
"loss": 1.9702026844024658,
"step": 4126
},
{
"epoch": 1.43935352267378,
"grad_norm": 0.2202550321817398,
"learning_rate": 3.3395599833956e-05,
"loss": 1.8255903720855713,
"step": 4128
},
{
"epoch": 1.4400509092177067,
"grad_norm": 0.22926436364650726,
"learning_rate": 3.335408883354089e-05,
"loss": 1.9207851886749268,
"step": 4130
},
{
"epoch": 1.4407482957616333,
"grad_norm": 0.22136305272579193,
"learning_rate": 3.331257783312578e-05,
"loss": 1.8368303775787354,
"step": 4132
},
{
"epoch": 1.44144568230556,
"grad_norm": 0.21406474709510803,
"learning_rate": 3.3271066832710665e-05,
"loss": 1.7291842699050903,
"step": 4134
},
{
"epoch": 1.4421430688494865,
"grad_norm": 0.2021942436695099,
"learning_rate": 3.3229555832295565e-05,
"loss": 2.00429105758667,
"step": 4136
},
{
"epoch": 1.442840455393413,
"grad_norm": 0.17133504152297974,
"learning_rate": 3.318804483188045e-05,
"loss": 1.8942899703979492,
"step": 4138
},
{
"epoch": 1.44353784193734,
"grad_norm": 0.23039887845516205,
"learning_rate": 3.314653383146534e-05,
"loss": 1.8624145984649658,
"step": 4140
},
{
"epoch": 1.4442352284812665,
"grad_norm": 0.19916106760501862,
"learning_rate": 3.310502283105023e-05,
"loss": 1.8991811275482178,
"step": 4142
},
{
"epoch": 1.444932615025193,
"grad_norm": 0.2166530042886734,
"learning_rate": 3.306351183063512e-05,
"loss": 1.8596470355987549,
"step": 4144
},
{
"epoch": 1.4456300015691197,
"grad_norm": 0.18740181624889374,
"learning_rate": 3.302200083022001e-05,
"loss": 1.948809266090393,
"step": 4146
},
{
"epoch": 1.4463273881130463,
"grad_norm": 0.17723046243190765,
"learning_rate": 3.29804898298049e-05,
"loss": 1.8506078720092773,
"step": 4148
},
{
"epoch": 1.447024774656973,
"grad_norm": 0.21443305909633636,
"learning_rate": 3.293897882938979e-05,
"loss": 1.7285115718841553,
"step": 4150
},
{
"epoch": 1.4477221612008997,
"grad_norm": 0.19111402332782745,
"learning_rate": 3.2897467828974684e-05,
"loss": 1.6482958793640137,
"step": 4152
},
{
"epoch": 1.4484195477448263,
"grad_norm": 0.3252273201942444,
"learning_rate": 3.285595682855957e-05,
"loss": 1.6977410316467285,
"step": 4154
},
{
"epoch": 1.4491169342887529,
"grad_norm": 0.2883060574531555,
"learning_rate": 3.281444582814446e-05,
"loss": 1.6457706689834595,
"step": 4156
},
{
"epoch": 1.4498143208326795,
"grad_norm": 0.235630065202713,
"learning_rate": 3.277293482772935e-05,
"loss": 1.8759286403656006,
"step": 4158
},
{
"epoch": 1.4505117073766063,
"grad_norm": 0.18711936473846436,
"learning_rate": 3.2731423827314236e-05,
"loss": 1.8516727685928345,
"step": 4160
},
{
"epoch": 1.4512090939205329,
"grad_norm": 0.22569718956947327,
"learning_rate": 3.268991282689913e-05,
"loss": 2.0689921379089355,
"step": 4162
},
{
"epoch": 1.4519064804644595,
"grad_norm": 0.18883809447288513,
"learning_rate": 3.264840182648402e-05,
"loss": 1.899646520614624,
"step": 4164
},
{
"epoch": 1.452603867008386,
"grad_norm": 0.21534138917922974,
"learning_rate": 3.260689082606891e-05,
"loss": 2.1046838760375977,
"step": 4166
},
{
"epoch": 1.4533012535523127,
"grad_norm": 0.27049872279167175,
"learning_rate": 3.25653798256538e-05,
"loss": 2.0120396614074707,
"step": 4168
},
{
"epoch": 1.4539986400962395,
"grad_norm": 0.23462948203086853,
"learning_rate": 3.252386882523869e-05,
"loss": 1.9374223947525024,
"step": 4170
},
{
"epoch": 1.4546960266401658,
"grad_norm": 0.24656078219413757,
"learning_rate": 3.2482357824823576e-05,
"loss": 1.8995258808135986,
"step": 4172
},
{
"epoch": 1.4553934131840927,
"grad_norm": 0.1956031173467636,
"learning_rate": 3.244084682440847e-05,
"loss": 1.9521234035491943,
"step": 4174
},
{
"epoch": 1.4560907997280192,
"grad_norm": 0.29042989015579224,
"learning_rate": 3.239933582399336e-05,
"loss": 1.9295008182525635,
"step": 4176
},
{
"epoch": 1.4567881862719458,
"grad_norm": 0.18132352828979492,
"learning_rate": 3.235782482357825e-05,
"loss": 1.949939250946045,
"step": 4178
},
{
"epoch": 1.4574855728158727,
"grad_norm": 0.2633827328681946,
"learning_rate": 3.231631382316314e-05,
"loss": 1.9462300539016724,
"step": 4180
},
{
"epoch": 1.458182959359799,
"grad_norm": 0.29535162448883057,
"learning_rate": 3.227480282274803e-05,
"loss": 1.8570556640625,
"step": 4182
},
{
"epoch": 1.4588803459037258,
"grad_norm": 0.19173215329647064,
"learning_rate": 3.2233291822332915e-05,
"loss": 2.079622745513916,
"step": 4184
},
{
"epoch": 1.4595777324476524,
"grad_norm": 0.22502847015857697,
"learning_rate": 3.219178082191781e-05,
"loss": 1.883758306503296,
"step": 4186
},
{
"epoch": 1.460275118991579,
"grad_norm": 0.22633635997772217,
"learning_rate": 3.21502698215027e-05,
"loss": 1.8361179828643799,
"step": 4188
},
{
"epoch": 1.4609725055355056,
"grad_norm": 0.20793241262435913,
"learning_rate": 3.2108758821087595e-05,
"loss": 2.172873020172119,
"step": 4190
},
{
"epoch": 1.4616698920794322,
"grad_norm": 0.22675487399101257,
"learning_rate": 3.206724782067248e-05,
"loss": 1.9477754831314087,
"step": 4192
},
{
"epoch": 1.462367278623359,
"grad_norm": 0.26028743386268616,
"learning_rate": 3.202573682025737e-05,
"loss": 1.8193738460540771,
"step": 4194
},
{
"epoch": 1.4630646651672856,
"grad_norm": 0.2475419044494629,
"learning_rate": 3.198422581984226e-05,
"loss": 1.974073886871338,
"step": 4196
},
{
"epoch": 1.4637620517112122,
"grad_norm": 0.2499392181634903,
"learning_rate": 3.194271481942715e-05,
"loss": 1.8304438591003418,
"step": 4198
},
{
"epoch": 1.4644594382551388,
"grad_norm": 0.237641379237175,
"learning_rate": 3.190120381901204e-05,
"loss": 1.8094310760498047,
"step": 4200
},
{
"epoch": 1.4651568247990654,
"grad_norm": 0.3710329234600067,
"learning_rate": 3.1859692818596934e-05,
"loss": 1.8436793088912964,
"step": 4202
},
{
"epoch": 1.4658542113429922,
"grad_norm": 0.17323395609855652,
"learning_rate": 3.181818181818182e-05,
"loss": 1.7896754741668701,
"step": 4204
},
{
"epoch": 1.4665515978869188,
"grad_norm": 0.19435246288776398,
"learning_rate": 3.177667081776671e-05,
"loss": 2.132152795791626,
"step": 4206
},
{
"epoch": 1.4672489844308454,
"grad_norm": 0.2881922125816345,
"learning_rate": 3.17351598173516e-05,
"loss": 1.7988765239715576,
"step": 4208
},
{
"epoch": 1.467946370974772,
"grad_norm": 0.19056060910224915,
"learning_rate": 3.1693648816936487e-05,
"loss": 1.7471990585327148,
"step": 4210
},
{
"epoch": 1.4686437575186986,
"grad_norm": 0.20850984752178192,
"learning_rate": 3.165213781652138e-05,
"loss": 1.7588502168655396,
"step": 4212
},
{
"epoch": 1.4693411440626254,
"grad_norm": 0.2419462502002716,
"learning_rate": 3.161062681610627e-05,
"loss": 1.7379381656646729,
"step": 4214
},
{
"epoch": 1.470038530606552,
"grad_norm": 0.27557116746902466,
"learning_rate": 3.156911581569116e-05,
"loss": 1.7018725872039795,
"step": 4216
},
{
"epoch": 1.4707359171504786,
"grad_norm": 0.21285809576511383,
"learning_rate": 3.152760481527605e-05,
"loss": 1.9607921838760376,
"step": 4218
},
{
"epoch": 1.4714333036944052,
"grad_norm": 0.233015775680542,
"learning_rate": 3.148609381486094e-05,
"loss": 1.8311192989349365,
"step": 4220
},
{
"epoch": 1.4721306902383318,
"grad_norm": 0.19445312023162842,
"learning_rate": 3.1444582814445826e-05,
"loss": 2.0046186447143555,
"step": 4222
},
{
"epoch": 1.4728280767822586,
"grad_norm": 0.25397178530693054,
"learning_rate": 3.140307181403072e-05,
"loss": 1.7237898111343384,
"step": 4224
},
{
"epoch": 1.4735254633261852,
"grad_norm": 0.17368128895759583,
"learning_rate": 3.136156081361561e-05,
"loss": 1.955993890762329,
"step": 4226
},
{
"epoch": 1.4742228498701118,
"grad_norm": 0.2297562062740326,
"learning_rate": 3.13200498132005e-05,
"loss": 1.8776400089263916,
"step": 4228
},
{
"epoch": 1.4749202364140384,
"grad_norm": 0.2735278308391571,
"learning_rate": 3.127853881278539e-05,
"loss": 1.8230985403060913,
"step": 4230
},
{
"epoch": 1.475617622957965,
"grad_norm": 0.2162596881389618,
"learning_rate": 3.123702781237028e-05,
"loss": 1.7406185865402222,
"step": 4232
},
{
"epoch": 1.4763150095018918,
"grad_norm": 0.26232606172561646,
"learning_rate": 3.119551681195517e-05,
"loss": 2.011214256286621,
"step": 4234
},
{
"epoch": 1.4770123960458184,
"grad_norm": 0.25555673241615295,
"learning_rate": 3.115400581154006e-05,
"loss": 1.908860683441162,
"step": 4236
},
{
"epoch": 1.477709782589745,
"grad_norm": 0.27135029435157776,
"learning_rate": 3.1112494811124945e-05,
"loss": 1.7173230648040771,
"step": 4238
},
{
"epoch": 1.4784071691336715,
"grad_norm": 0.18783152103424072,
"learning_rate": 3.1070983810709845e-05,
"loss": 1.9788048267364502,
"step": 4240
},
{
"epoch": 1.4791045556775981,
"grad_norm": 0.22805720567703247,
"learning_rate": 3.102947281029473e-05,
"loss": 2.030653476715088,
"step": 4242
},
{
"epoch": 1.479801942221525,
"grad_norm": 0.1582372486591339,
"learning_rate": 3.098796180987962e-05,
"loss": 1.6906356811523438,
"step": 4244
},
{
"epoch": 1.4804993287654515,
"grad_norm": 0.19490480422973633,
"learning_rate": 3.094645080946451e-05,
"loss": 1.8677330017089844,
"step": 4246
},
{
"epoch": 1.4811967153093781,
"grad_norm": 0.22457078099250793,
"learning_rate": 3.09049398090494e-05,
"loss": 2.066915512084961,
"step": 4248
},
{
"epoch": 1.4818941018533047,
"grad_norm": 0.19483765959739685,
"learning_rate": 3.0863428808634284e-05,
"loss": 1.9681627750396729,
"step": 4250
},
{
"epoch": 1.4825914883972313,
"grad_norm": 0.18761757016181946,
"learning_rate": 3.082191780821918e-05,
"loss": 1.7936248779296875,
"step": 4252
},
{
"epoch": 1.4832888749411581,
"grad_norm": 0.24503754079341888,
"learning_rate": 3.078040680780407e-05,
"loss": 2.069383144378662,
"step": 4254
},
{
"epoch": 1.4839862614850847,
"grad_norm": 0.35354992747306824,
"learning_rate": 3.0738895807388964e-05,
"loss": 1.8734214305877686,
"step": 4256
},
{
"epoch": 1.4846836480290113,
"grad_norm": 0.2514292001724243,
"learning_rate": 3.069738480697385e-05,
"loss": 1.918771743774414,
"step": 4258
},
{
"epoch": 1.485381034572938,
"grad_norm": 0.1698964387178421,
"learning_rate": 3.065587380655874e-05,
"loss": 1.973615288734436,
"step": 4260
},
{
"epoch": 1.4860784211168645,
"grad_norm": 0.20615416765213013,
"learning_rate": 3.061436280614363e-05,
"loss": 1.7398695945739746,
"step": 4262
},
{
"epoch": 1.4867758076607913,
"grad_norm": 0.27245327830314636,
"learning_rate": 3.0572851805728516e-05,
"loss": 1.8311172723770142,
"step": 4264
},
{
"epoch": 1.4874731942047177,
"grad_norm": 0.28381961584091187,
"learning_rate": 3.053134080531341e-05,
"loss": 2.0440845489501953,
"step": 4266
},
{
"epoch": 1.4881705807486445,
"grad_norm": 0.22662436962127686,
"learning_rate": 3.04898298048983e-05,
"loss": 1.9301950931549072,
"step": 4268
},
{
"epoch": 1.488867967292571,
"grad_norm": 0.18592658638954163,
"learning_rate": 3.044831880448319e-05,
"loss": 1.7836358547210693,
"step": 4270
},
{
"epoch": 1.4895653538364977,
"grad_norm": 0.21467696130275726,
"learning_rate": 3.0406807804068076e-05,
"loss": 1.9252032041549683,
"step": 4272
},
{
"epoch": 1.4902627403804243,
"grad_norm": 0.22956031560897827,
"learning_rate": 3.036529680365297e-05,
"loss": 1.911210298538208,
"step": 4274
},
{
"epoch": 1.4909601269243509,
"grad_norm": 0.19773449003696442,
"learning_rate": 3.032378580323786e-05,
"loss": 1.9682331085205078,
"step": 4276
},
{
"epoch": 1.4916575134682777,
"grad_norm": 0.19668564200401306,
"learning_rate": 3.0282274802822752e-05,
"loss": 2.0410146713256836,
"step": 4278
},
{
"epoch": 1.4923549000122043,
"grad_norm": 0.2678745687007904,
"learning_rate": 3.024076380240764e-05,
"loss": 1.948699712753296,
"step": 4280
},
{
"epoch": 1.4930522865561309,
"grad_norm": 0.15882091224193573,
"learning_rate": 3.019925280199253e-05,
"loss": 2.104851245880127,
"step": 4282
},
{
"epoch": 1.4937496731000575,
"grad_norm": 0.25176993012428284,
"learning_rate": 3.0157741801577422e-05,
"loss": 1.7860289812088013,
"step": 4284
},
{
"epoch": 1.494447059643984,
"grad_norm": 0.21591933071613312,
"learning_rate": 3.011623080116231e-05,
"loss": 1.8135545253753662,
"step": 4286
},
{
"epoch": 1.4951444461879109,
"grad_norm": 0.23614440858364105,
"learning_rate": 3.00747198007472e-05,
"loss": 1.829087734222412,
"step": 4288
},
{
"epoch": 1.4958418327318375,
"grad_norm": 0.21727608144283295,
"learning_rate": 3.003320880033209e-05,
"loss": 1.7977166175842285,
"step": 4290
},
{
"epoch": 1.496539219275764,
"grad_norm": 0.19061827659606934,
"learning_rate": 2.9991697799916978e-05,
"loss": 1.9947948455810547,
"step": 4292
},
{
"epoch": 1.4972366058196906,
"grad_norm": 0.24229376018047333,
"learning_rate": 2.9950186799501868e-05,
"loss": 1.9775480031967163,
"step": 4294
},
{
"epoch": 1.4979339923636172,
"grad_norm": 0.17827463150024414,
"learning_rate": 2.990867579908676e-05,
"loss": 1.904175043106079,
"step": 4296
},
{
"epoch": 1.498631378907544,
"grad_norm": 0.24656245112419128,
"learning_rate": 2.9867164798671648e-05,
"loss": 1.9311580657958984,
"step": 4298
},
{
"epoch": 1.4993287654514706,
"grad_norm": 0.2341352105140686,
"learning_rate": 2.982565379825654e-05,
"loss": 1.8169183731079102,
"step": 4300
},
{
"epoch": 1.5000261519953972,
"grad_norm": 0.21976277232170105,
"learning_rate": 2.978414279784143e-05,
"loss": 1.9069582223892212,
"step": 4302
},
{
"epoch": 1.5007235385393238,
"grad_norm": 0.18122829496860504,
"learning_rate": 2.9742631797426317e-05,
"loss": 1.9984383583068848,
"step": 4304
},
{
"epoch": 1.5014209250832504,
"grad_norm": 0.527238130569458,
"learning_rate": 2.970112079701121e-05,
"loss": 1.7675752639770508,
"step": 4306
},
{
"epoch": 1.5021183116271772,
"grad_norm": 0.23009146749973297,
"learning_rate": 2.96596097965961e-05,
"loss": 1.8511323928833008,
"step": 4308
},
{
"epoch": 1.5028156981711038,
"grad_norm": 0.2193315178155899,
"learning_rate": 2.9618098796180987e-05,
"loss": 1.7352030277252197,
"step": 4310
},
{
"epoch": 1.5035130847150304,
"grad_norm": 0.2285013347864151,
"learning_rate": 2.957658779576588e-05,
"loss": 2.0133321285247803,
"step": 4312
},
{
"epoch": 1.504210471258957,
"grad_norm": 0.1797766536474228,
"learning_rate": 2.953507679535077e-05,
"loss": 1.6843817234039307,
"step": 4314
},
{
"epoch": 1.5049078578028836,
"grad_norm": 0.20497511327266693,
"learning_rate": 2.9493565794935657e-05,
"loss": 1.9545986652374268,
"step": 4316
},
{
"epoch": 1.5056052443468104,
"grad_norm": 0.3286879360675812,
"learning_rate": 2.945205479452055e-05,
"loss": 1.9560670852661133,
"step": 4318
},
{
"epoch": 1.506302630890737,
"grad_norm": 0.2364288568496704,
"learning_rate": 2.941054379410544e-05,
"loss": 1.679736852645874,
"step": 4320
},
{
"epoch": 1.5070000174346636,
"grad_norm": 0.20072239637374878,
"learning_rate": 2.9369032793690333e-05,
"loss": 1.8908525705337524,
"step": 4322
},
{
"epoch": 1.5076974039785902,
"grad_norm": 0.32942745089530945,
"learning_rate": 2.932752179327522e-05,
"loss": 1.7327306270599365,
"step": 4324
},
{
"epoch": 1.5083947905225168,
"grad_norm": 0.2948322594165802,
"learning_rate": 2.928601079286011e-05,
"loss": 1.835234522819519,
"step": 4326
},
{
"epoch": 1.5090921770664436,
"grad_norm": 0.205282524228096,
"learning_rate": 2.9244499792445003e-05,
"loss": 1.8638992309570312,
"step": 4328
},
{
"epoch": 1.50978956361037,
"grad_norm": 0.2971500754356384,
"learning_rate": 2.920298879202989e-05,
"loss": 1.9175176620483398,
"step": 4330
},
{
"epoch": 1.5104869501542968,
"grad_norm": 0.18545030057430267,
"learning_rate": 2.916147779161478e-05,
"loss": 1.9995113611221313,
"step": 4332
},
{
"epoch": 1.5111843366982234,
"grad_norm": 0.1809265911579132,
"learning_rate": 2.9119966791199672e-05,
"loss": 1.9885342121124268,
"step": 4334
},
{
"epoch": 1.51188172324215,
"grad_norm": 0.24308079481124878,
"learning_rate": 2.907845579078456e-05,
"loss": 1.9155973196029663,
"step": 4336
},
{
"epoch": 1.5125791097860768,
"grad_norm": 0.21333220601081848,
"learning_rate": 2.903694479036945e-05,
"loss": 1.998091697692871,
"step": 4338
},
{
"epoch": 1.5132764963300032,
"grad_norm": 0.23621611297130585,
"learning_rate": 2.8995433789954342e-05,
"loss": 1.6689598560333252,
"step": 4340
},
{
"epoch": 1.51397388287393,
"grad_norm": 0.19262677431106567,
"learning_rate": 2.8953922789539228e-05,
"loss": 2.021580696105957,
"step": 4342
},
{
"epoch": 1.5146712694178566,
"grad_norm": 0.21616120636463165,
"learning_rate": 2.891241178912412e-05,
"loss": 1.8001134395599365,
"step": 4344
},
{
"epoch": 1.5153686559617832,
"grad_norm": 0.24169668555259705,
"learning_rate": 2.887090078870901e-05,
"loss": 1.6586966514587402,
"step": 4346
},
{
"epoch": 1.51606604250571,
"grad_norm": 0.24617350101470947,
"learning_rate": 2.8829389788293898e-05,
"loss": 1.7206151485443115,
"step": 4348
},
{
"epoch": 1.5167634290496363,
"grad_norm": 0.16805648803710938,
"learning_rate": 2.878787878787879e-05,
"loss": 1.9847954511642456,
"step": 4350
},
{
"epoch": 1.5174608155935632,
"grad_norm": 0.17351919412612915,
"learning_rate": 2.874636778746368e-05,
"loss": 1.878028154373169,
"step": 4352
},
{
"epoch": 1.5181582021374898,
"grad_norm": 0.19319580495357513,
"learning_rate": 2.8704856787048567e-05,
"loss": 1.91669499874115,
"step": 4354
},
{
"epoch": 1.5188555886814163,
"grad_norm": 0.18220485746860504,
"learning_rate": 2.866334578663346e-05,
"loss": 1.76810884475708,
"step": 4356
},
{
"epoch": 1.5195529752253432,
"grad_norm": 0.26242494583129883,
"learning_rate": 2.862183478621835e-05,
"loss": 1.7875232696533203,
"step": 4358
},
{
"epoch": 1.5202503617692695,
"grad_norm": 0.22367031872272491,
"learning_rate": 2.8580323785803237e-05,
"loss": 1.9266059398651123,
"step": 4360
},
{
"epoch": 1.5209477483131963,
"grad_norm": 0.3521316349506378,
"learning_rate": 2.853881278538813e-05,
"loss": 1.8663140535354614,
"step": 4362
},
{
"epoch": 1.521645134857123,
"grad_norm": 0.21653883159160614,
"learning_rate": 2.8497301784973017e-05,
"loss": 1.9207054376602173,
"step": 4364
},
{
"epoch": 1.5223425214010495,
"grad_norm": 0.19742324948310852,
"learning_rate": 2.8455790784557913e-05,
"loss": 1.9935832023620605,
"step": 4366
},
{
"epoch": 1.5230399079449763,
"grad_norm": 0.23529601097106934,
"learning_rate": 2.84142797841428e-05,
"loss": 1.8649508953094482,
"step": 4368
},
{
"epoch": 1.5237372944889027,
"grad_norm": 0.17153075337409973,
"learning_rate": 2.8372768783727686e-05,
"loss": 1.8191823959350586,
"step": 4370
},
{
"epoch": 1.5244346810328295,
"grad_norm": 0.24875670671463013,
"learning_rate": 2.8331257783312583e-05,
"loss": 1.959450364112854,
"step": 4372
},
{
"epoch": 1.5251320675767561,
"grad_norm": 0.23544973134994507,
"learning_rate": 2.828974678289747e-05,
"loss": 2.0190210342407227,
"step": 4374
},
{
"epoch": 1.5258294541206827,
"grad_norm": 0.18131598830223083,
"learning_rate": 2.8248235782482356e-05,
"loss": 1.9842783212661743,
"step": 4376
},
{
"epoch": 1.5265268406646095,
"grad_norm": 0.1840776801109314,
"learning_rate": 2.820672478206725e-05,
"loss": 2.0165488719940186,
"step": 4378
},
{
"epoch": 1.527224227208536,
"grad_norm": 0.2549632489681244,
"learning_rate": 2.816521378165214e-05,
"loss": 1.9990711212158203,
"step": 4380
},
{
"epoch": 1.5279216137524627,
"grad_norm": 0.20285262167453766,
"learning_rate": 2.8123702781237026e-05,
"loss": 1.9805941581726074,
"step": 4382
},
{
"epoch": 1.5286190002963893,
"grad_norm": 0.2610595226287842,
"learning_rate": 2.808219178082192e-05,
"loss": 1.9052305221557617,
"step": 4384
},
{
"epoch": 1.529316386840316,
"grad_norm": 0.19016319513320923,
"learning_rate": 2.804068078040681e-05,
"loss": 1.8738257884979248,
"step": 4386
},
{
"epoch": 1.5300137733842427,
"grad_norm": 0.24855868518352509,
"learning_rate": 2.7999169779991702e-05,
"loss": 1.9404520988464355,
"step": 4388
},
{
"epoch": 1.530711159928169,
"grad_norm": 0.24818985164165497,
"learning_rate": 2.795765877957659e-05,
"loss": 1.8301239013671875,
"step": 4390
},
{
"epoch": 1.531408546472096,
"grad_norm": 0.2025291919708252,
"learning_rate": 2.791614777916148e-05,
"loss": 1.9974764585494995,
"step": 4392
},
{
"epoch": 1.5321059330160225,
"grad_norm": 0.23878057301044464,
"learning_rate": 2.787463677874637e-05,
"loss": 1.8272838592529297,
"step": 4394
},
{
"epoch": 1.532803319559949,
"grad_norm": 0.16763503849506378,
"learning_rate": 2.7833125778331258e-05,
"loss": 1.6806379556655884,
"step": 4396
},
{
"epoch": 1.5335007061038757,
"grad_norm": 0.1883159577846527,
"learning_rate": 2.7791614777916148e-05,
"loss": 1.9163085222244263,
"step": 4398
},
{
"epoch": 1.5341980926478023,
"grad_norm": 0.22081787884235382,
"learning_rate": 2.775010377750104e-05,
"loss": 1.7372932434082031,
"step": 4400
},
{
"epoch": 1.534895479191729,
"grad_norm": 0.1913776695728302,
"learning_rate": 2.7708592777085928e-05,
"loss": 1.8430018424987793,
"step": 4402
},
{
"epoch": 1.5355928657356557,
"grad_norm": 0.17796553671360016,
"learning_rate": 2.7667081776670818e-05,
"loss": 1.8803733587265015,
"step": 4404
},
{
"epoch": 1.5362902522795823,
"grad_norm": 0.2805418074131012,
"learning_rate": 2.762557077625571e-05,
"loss": 2.04244327545166,
"step": 4406
},
{
"epoch": 1.5369876388235089,
"grad_norm": 0.18227362632751465,
"learning_rate": 2.7584059775840597e-05,
"loss": 1.9621096849441528,
"step": 4408
},
{
"epoch": 1.5376850253674355,
"grad_norm": 0.24022343754768372,
"learning_rate": 2.754254877542549e-05,
"loss": 1.6480050086975098,
"step": 4410
},
{
"epoch": 1.5383824119113623,
"grad_norm": 0.1897001564502716,
"learning_rate": 2.750103777501038e-05,
"loss": 1.7422631978988647,
"step": 4412
},
{
"epoch": 1.5390797984552886,
"grad_norm": 0.17075501382350922,
"learning_rate": 2.7459526774595267e-05,
"loss": 1.953518033027649,
"step": 4414
},
{
"epoch": 1.5397771849992155,
"grad_norm": 0.285891056060791,
"learning_rate": 2.741801577418016e-05,
"loss": 1.8298866748809814,
"step": 4416
},
{
"epoch": 1.540474571543142,
"grad_norm": 0.2010376900434494,
"learning_rate": 2.737650477376505e-05,
"loss": 1.6861653327941895,
"step": 4418
},
{
"epoch": 1.5411719580870686,
"grad_norm": 0.28427520394325256,
"learning_rate": 2.7334993773349937e-05,
"loss": 1.6987539529800415,
"step": 4420
},
{
"epoch": 1.5418693446309955,
"grad_norm": 0.26670607924461365,
"learning_rate": 2.729348277293483e-05,
"loss": 1.6583788394927979,
"step": 4422
},
{
"epoch": 1.5425667311749218,
"grad_norm": 0.21028931438922882,
"learning_rate": 2.725197177251972e-05,
"loss": 2.1383423805236816,
"step": 4424
},
{
"epoch": 1.5432641177188486,
"grad_norm": 0.19236426055431366,
"learning_rate": 2.7210460772104606e-05,
"loss": 1.8114062547683716,
"step": 4426
},
{
"epoch": 1.5439615042627752,
"grad_norm": 0.1726246178150177,
"learning_rate": 2.71689497716895e-05,
"loss": 1.966844081878662,
"step": 4428
},
{
"epoch": 1.5446588908067018,
"grad_norm": 0.18041053414344788,
"learning_rate": 2.712743877127439e-05,
"loss": 2.085667133331299,
"step": 4430
},
{
"epoch": 1.5453562773506286,
"grad_norm": 0.22201724350452423,
"learning_rate": 2.7085927770859276e-05,
"loss": 1.876138687133789,
"step": 4432
},
{
"epoch": 1.546053663894555,
"grad_norm": 0.2123415172100067,
"learning_rate": 2.704441677044417e-05,
"loss": 1.756211757659912,
"step": 4434
},
{
"epoch": 1.5467510504384818,
"grad_norm": 0.2050902545452118,
"learning_rate": 2.700290577002906e-05,
"loss": 1.9153730869293213,
"step": 4436
},
{
"epoch": 1.5474484369824084,
"grad_norm": 0.2251085489988327,
"learning_rate": 2.6961394769613952e-05,
"loss": 1.8800716400146484,
"step": 4438
},
{
"epoch": 1.548145823526335,
"grad_norm": 0.21331456303596497,
"learning_rate": 2.691988376919884e-05,
"loss": 1.902909278869629,
"step": 4440
},
{
"epoch": 1.5488432100702618,
"grad_norm": 0.17269904911518097,
"learning_rate": 2.687837276878373e-05,
"loss": 1.8585556745529175,
"step": 4442
},
{
"epoch": 1.5495405966141882,
"grad_norm": 0.2580990791320801,
"learning_rate": 2.6836861768368622e-05,
"loss": 1.897294044494629,
"step": 4444
},
{
"epoch": 1.550237983158115,
"grad_norm": 0.20308732986450195,
"learning_rate": 2.679535076795351e-05,
"loss": 1.9232985973358154,
"step": 4446
},
{
"epoch": 1.5509353697020416,
"grad_norm": 0.24419395625591278,
"learning_rate": 2.6753839767538398e-05,
"loss": 1.9122428894042969,
"step": 4448
},
{
"epoch": 1.5516327562459682,
"grad_norm": 0.17414434254169464,
"learning_rate": 2.671232876712329e-05,
"loss": 1.7522237300872803,
"step": 4450
},
{
"epoch": 1.552330142789895,
"grad_norm": 0.24114203453063965,
"learning_rate": 2.6670817766708178e-05,
"loss": 1.7971198558807373,
"step": 4452
},
{
"epoch": 1.5530275293338214,
"grad_norm": 0.17447638511657715,
"learning_rate": 2.6629306766293068e-05,
"loss": 1.81462824344635,
"step": 4454
},
{
"epoch": 1.5537249158777482,
"grad_norm": 0.21867063641548157,
"learning_rate": 2.658779576587796e-05,
"loss": 1.854804277420044,
"step": 4456
},
{
"epoch": 1.5544223024216748,
"grad_norm": 0.16868700087070465,
"learning_rate": 2.6546284765462848e-05,
"loss": 2.0059430599212646,
"step": 4458
},
{
"epoch": 1.5551196889656014,
"grad_norm": 0.2346510887145996,
"learning_rate": 2.650477376504774e-05,
"loss": 1.4958382844924927,
"step": 4460
},
{
"epoch": 1.5558170755095282,
"grad_norm": 0.26675206422805786,
"learning_rate": 2.646326276463263e-05,
"loss": 1.9033410549163818,
"step": 4462
},
{
"epoch": 1.5565144620534546,
"grad_norm": 0.18497225642204285,
"learning_rate": 2.6421751764217517e-05,
"loss": 1.9528168439865112,
"step": 4464
},
{
"epoch": 1.5572118485973814,
"grad_norm": 0.20328418910503387,
"learning_rate": 2.638024076380241e-05,
"loss": 1.83001708984375,
"step": 4466
},
{
"epoch": 1.557909235141308,
"grad_norm": 0.21240051090717316,
"learning_rate": 2.63387297633873e-05,
"loss": 1.8499531745910645,
"step": 4468
},
{
"epoch": 1.5586066216852346,
"grad_norm": 0.24162444472312927,
"learning_rate": 2.6297218762972187e-05,
"loss": 1.6314120292663574,
"step": 4470
},
{
"epoch": 1.5593040082291614,
"grad_norm": 0.27374324202537537,
"learning_rate": 2.625570776255708e-05,
"loss": 1.7221617698669434,
"step": 4472
},
{
"epoch": 1.5600013947730877,
"grad_norm": 0.25410524010658264,
"learning_rate": 2.6214196762141967e-05,
"loss": 1.7684450149536133,
"step": 4474
},
{
"epoch": 1.5606987813170146,
"grad_norm": 0.2422337681055069,
"learning_rate": 2.6172685761726856e-05,
"loss": 1.8309746980667114,
"step": 4476
},
{
"epoch": 1.5613961678609412,
"grad_norm": 0.23445764183998108,
"learning_rate": 2.613117476131175e-05,
"loss": 2.1427254676818848,
"step": 4478
},
{
"epoch": 1.5620935544048677,
"grad_norm": 0.1943419724702835,
"learning_rate": 2.6089663760896636e-05,
"loss": 1.7775300741195679,
"step": 4480
},
{
"epoch": 1.5627909409487943,
"grad_norm": 0.20946148037910461,
"learning_rate": 2.6048152760481533e-05,
"loss": 1.8440033197402954,
"step": 4482
},
{
"epoch": 1.563488327492721,
"grad_norm": 0.19684435427188873,
"learning_rate": 2.600664176006642e-05,
"loss": 1.8125154972076416,
"step": 4484
},
{
"epoch": 1.5641857140366477,
"grad_norm": 0.26996201276779175,
"learning_rate": 2.5965130759651306e-05,
"loss": 1.7742915153503418,
"step": 4486
},
{
"epoch": 1.5648831005805743,
"grad_norm": 0.174392968416214,
"learning_rate": 2.59236197592362e-05,
"loss": 1.9183518886566162,
"step": 4488
},
{
"epoch": 1.565580487124501,
"grad_norm": 0.17263713479042053,
"learning_rate": 2.588210875882109e-05,
"loss": 1.842463731765747,
"step": 4490
},
{
"epoch": 1.5662778736684275,
"grad_norm": 0.2301662117242813,
"learning_rate": 2.5840597758405975e-05,
"loss": 1.9093573093414307,
"step": 4492
},
{
"epoch": 1.5669752602123541,
"grad_norm": 0.1826668381690979,
"learning_rate": 2.579908675799087e-05,
"loss": 1.9088093042373657,
"step": 4494
},
{
"epoch": 1.567672646756281,
"grad_norm": 0.18410944938659668,
"learning_rate": 2.575757575757576e-05,
"loss": 1.831756830215454,
"step": 4496
},
{
"epoch": 1.5683700333002075,
"grad_norm": 0.22626711428165436,
"learning_rate": 2.5716064757160645e-05,
"loss": 2.102736711502075,
"step": 4498
},
{
"epoch": 1.5690674198441341,
"grad_norm": 0.24183250963687897,
"learning_rate": 2.5674553756745538e-05,
"loss": 1.976166009902954,
"step": 4500
},
{
"epoch": 1.5697648063880607,
"grad_norm": 0.24379684031009674,
"learning_rate": 2.5633042756330428e-05,
"loss": 1.8924784660339355,
"step": 4502
},
{
"epoch": 1.5704621929319873,
"grad_norm": 0.17403602600097656,
"learning_rate": 2.559153175591532e-05,
"loss": 2.0470175743103027,
"step": 4504
},
{
"epoch": 1.5711595794759141,
"grad_norm": 0.1863819658756256,
"learning_rate": 2.5550020755500208e-05,
"loss": 2.1825854778289795,
"step": 4506
},
{
"epoch": 1.5718569660198405,
"grad_norm": 0.16507472097873688,
"learning_rate": 2.5508509755085098e-05,
"loss": 1.951021671295166,
"step": 4508
},
{
"epoch": 1.5725543525637673,
"grad_norm": 0.22446586191654205,
"learning_rate": 2.546699875466999e-05,
"loss": 1.9840681552886963,
"step": 4510
},
{
"epoch": 1.573251739107694,
"grad_norm": 0.23697252571582794,
"learning_rate": 2.5425487754254877e-05,
"loss": 2.0239810943603516,
"step": 4512
},
{
"epoch": 1.5739491256516205,
"grad_norm": 0.22238145768642426,
"learning_rate": 2.5383976753839767e-05,
"loss": 1.808410406112671,
"step": 4514
},
{
"epoch": 1.5746465121955473,
"grad_norm": 0.31593209505081177,
"learning_rate": 2.534246575342466e-05,
"loss": 1.7202047109603882,
"step": 4516
},
{
"epoch": 1.5753438987394737,
"grad_norm": 0.2227022647857666,
"learning_rate": 2.5300954753009547e-05,
"loss": 2.1358001232147217,
"step": 4518
},
{
"epoch": 1.5760412852834005,
"grad_norm": 0.18909968435764313,
"learning_rate": 2.5259443752594437e-05,
"loss": 1.8523045778274536,
"step": 4520
},
{
"epoch": 1.576738671827327,
"grad_norm": 0.2606893479824066,
"learning_rate": 2.521793275217933e-05,
"loss": 1.898789644241333,
"step": 4522
},
{
"epoch": 1.5774360583712537,
"grad_norm": 0.21320763230323792,
"learning_rate": 2.5176421751764217e-05,
"loss": 1.7763218879699707,
"step": 4524
},
{
"epoch": 1.5781334449151805,
"grad_norm": 0.20257368683815002,
"learning_rate": 2.513491075134911e-05,
"loss": 1.7035415172576904,
"step": 4526
},
{
"epoch": 1.5788308314591069,
"grad_norm": 0.24113517999649048,
"learning_rate": 2.5093399750934e-05,
"loss": 1.7718031406402588,
"step": 4528
},
{
"epoch": 1.5795282180030337,
"grad_norm": 0.23414462804794312,
"learning_rate": 2.5051888750518886e-05,
"loss": 1.9617892503738403,
"step": 4530
},
{
"epoch": 1.5802256045469603,
"grad_norm": 0.20183584094047546,
"learning_rate": 2.501037775010378e-05,
"loss": 1.7238619327545166,
"step": 4532
},
{
"epoch": 1.5809229910908869,
"grad_norm": 0.21193481981754303,
"learning_rate": 2.496886674968867e-05,
"loss": 1.9761714935302734,
"step": 4534
},
{
"epoch": 1.5816203776348137,
"grad_norm": 0.23748309910297394,
"learning_rate": 2.492735574927356e-05,
"loss": 1.8902467489242554,
"step": 4536
},
{
"epoch": 1.58231776417874,
"grad_norm": 0.20187820494174957,
"learning_rate": 2.4885844748858446e-05,
"loss": 1.877758264541626,
"step": 4538
},
{
"epoch": 1.5830151507226669,
"grad_norm": 0.23040840029716492,
"learning_rate": 2.484433374844334e-05,
"loss": 1.8351686000823975,
"step": 4540
},
{
"epoch": 1.5837125372665934,
"grad_norm": 0.21899482607841492,
"learning_rate": 2.480282274802823e-05,
"loss": 1.979339599609375,
"step": 4542
},
{
"epoch": 1.58440992381052,
"grad_norm": 0.2788389325141907,
"learning_rate": 2.476131174761312e-05,
"loss": 1.9334053993225098,
"step": 4544
},
{
"epoch": 1.5851073103544469,
"grad_norm": 0.15392333269119263,
"learning_rate": 2.471980074719801e-05,
"loss": 1.9474034309387207,
"step": 4546
},
{
"epoch": 1.5858046968983732,
"grad_norm": 0.24061237275600433,
"learning_rate": 2.46782897467829e-05,
"loss": 1.825833797454834,
"step": 4548
},
{
"epoch": 1.5865020834423,
"grad_norm": 0.22537417709827423,
"learning_rate": 2.463677874636779e-05,
"loss": 1.8872768878936768,
"step": 4550
},
{
"epoch": 1.5871994699862266,
"grad_norm": 0.1721351444721222,
"learning_rate": 2.4595267745952678e-05,
"loss": 1.86381995677948,
"step": 4552
},
{
"epoch": 1.5878968565301532,
"grad_norm": 0.2710893750190735,
"learning_rate": 2.4553756745537568e-05,
"loss": 1.5904405117034912,
"step": 4554
},
{
"epoch": 1.58859424307408,
"grad_norm": 0.24607409536838531,
"learning_rate": 2.4512245745122458e-05,
"loss": 1.835451364517212,
"step": 4556
},
{
"epoch": 1.5892916296180064,
"grad_norm": 0.1593160331249237,
"learning_rate": 2.4470734744707348e-05,
"loss": 1.8716810941696167,
"step": 4558
},
{
"epoch": 1.5899890161619332,
"grad_norm": 0.1712384670972824,
"learning_rate": 2.4429223744292238e-05,
"loss": 1.8859171867370605,
"step": 4560
},
{
"epoch": 1.5906864027058598,
"grad_norm": 0.1740417182445526,
"learning_rate": 2.4387712743877128e-05,
"loss": 1.9691619873046875,
"step": 4562
},
{
"epoch": 1.5913837892497864,
"grad_norm": 0.2035098373889923,
"learning_rate": 2.4346201743462018e-05,
"loss": 1.9540915489196777,
"step": 4564
},
{
"epoch": 1.5920811757937132,
"grad_norm": 0.17674970626831055,
"learning_rate": 2.430469074304691e-05,
"loss": 2.0212907791137695,
"step": 4566
},
{
"epoch": 1.5927785623376396,
"grad_norm": 0.24654126167297363,
"learning_rate": 2.4263179742631797e-05,
"loss": 1.8362865447998047,
"step": 4568
},
{
"epoch": 1.5934759488815664,
"grad_norm": 0.2556804418563843,
"learning_rate": 2.4221668742216687e-05,
"loss": 1.67009437084198,
"step": 4570
},
{
"epoch": 1.594173335425493,
"grad_norm": 0.17135943472385406,
"learning_rate": 2.418015774180158e-05,
"loss": 1.947709560394287,
"step": 4572
},
{
"epoch": 1.5948707219694196,
"grad_norm": 0.24243444204330444,
"learning_rate": 2.413864674138647e-05,
"loss": 1.7617876529693604,
"step": 4574
},
{
"epoch": 1.5955681085133462,
"grad_norm": 0.33450421690940857,
"learning_rate": 2.4097135740971357e-05,
"loss": 1.7707524299621582,
"step": 4576
},
{
"epoch": 1.5962654950572728,
"grad_norm": 0.27597707509994507,
"learning_rate": 2.405562474055625e-05,
"loss": 1.9219856262207031,
"step": 4578
},
{
"epoch": 1.5969628816011996,
"grad_norm": 0.21139302849769592,
"learning_rate": 2.401411374014114e-05,
"loss": 2.03281831741333,
"step": 4580
},
{
"epoch": 1.5976602681451262,
"grad_norm": 0.22571788728237152,
"learning_rate": 2.3972602739726026e-05,
"loss": 2.055140733718872,
"step": 4582
},
{
"epoch": 1.5983576546890528,
"grad_norm": 0.24944792687892914,
"learning_rate": 2.393109173931092e-05,
"loss": 1.7604517936706543,
"step": 4584
},
{
"epoch": 1.5990550412329794,
"grad_norm": 0.22140969336032867,
"learning_rate": 2.388958073889581e-05,
"loss": 1.9736140966415405,
"step": 4586
},
{
"epoch": 1.599752427776906,
"grad_norm": 0.24355095624923706,
"learning_rate": 2.38480697384807e-05,
"loss": 1.6559598445892334,
"step": 4588
},
{
"epoch": 1.6004498143208328,
"grad_norm": 0.19067931175231934,
"learning_rate": 2.3806558738065586e-05,
"loss": 2.0552828311920166,
"step": 4590
},
{
"epoch": 1.6011472008647594,
"grad_norm": 0.23808273673057556,
"learning_rate": 2.376504773765048e-05,
"loss": 1.7038203477859497,
"step": 4592
},
{
"epoch": 1.601844587408686,
"grad_norm": 0.1991022229194641,
"learning_rate": 2.372353673723537e-05,
"loss": 1.6629961729049683,
"step": 4594
},
{
"epoch": 1.6025419739526126,
"grad_norm": 0.3089389503002167,
"learning_rate": 2.368202573682026e-05,
"loss": 1.8627169132232666,
"step": 4596
},
{
"epoch": 1.6032393604965391,
"grad_norm": 0.1592096984386444,
"learning_rate": 2.364051473640515e-05,
"loss": 1.7201133966445923,
"step": 4598
},
{
"epoch": 1.603936747040466,
"grad_norm": 0.2210923135280609,
"learning_rate": 2.359900373599004e-05,
"loss": 1.8774983882904053,
"step": 4600
},
{
"epoch": 1.6046341335843923,
"grad_norm": 0.23607036471366882,
"learning_rate": 2.355749273557493e-05,
"loss": 1.817946195602417,
"step": 4602
},
{
"epoch": 1.6053315201283191,
"grad_norm": 0.19019250571727753,
"learning_rate": 2.351598173515982e-05,
"loss": 1.9907363653182983,
"step": 4604
},
{
"epoch": 1.6060289066722457,
"grad_norm": 0.2500857412815094,
"learning_rate": 2.3474470734744708e-05,
"loss": 1.7164638042449951,
"step": 4606
},
{
"epoch": 1.6067262932161723,
"grad_norm": 0.20450207591056824,
"learning_rate": 2.3432959734329598e-05,
"loss": 1.8848949670791626,
"step": 4608
},
{
"epoch": 1.6074236797600991,
"grad_norm": 0.17902444303035736,
"learning_rate": 2.3391448733914488e-05,
"loss": 1.8511407375335693,
"step": 4610
},
{
"epoch": 1.6081210663040255,
"grad_norm": 0.31758981943130493,
"learning_rate": 2.3349937733499378e-05,
"loss": 1.874647855758667,
"step": 4612
},
{
"epoch": 1.6088184528479523,
"grad_norm": 0.1628735065460205,
"learning_rate": 2.3308426733084268e-05,
"loss": 1.9878525733947754,
"step": 4614
},
{
"epoch": 1.609515839391879,
"grad_norm": 0.2540745735168457,
"learning_rate": 2.3266915732669158e-05,
"loss": 1.712442398071289,
"step": 4616
},
{
"epoch": 1.6102132259358055,
"grad_norm": 0.23754636943340302,
"learning_rate": 2.322540473225405e-05,
"loss": 1.831302523612976,
"step": 4618
},
{
"epoch": 1.6109106124797323,
"grad_norm": 0.24721118807792664,
"learning_rate": 2.3183893731838937e-05,
"loss": 1.9953370094299316,
"step": 4620
},
{
"epoch": 1.6116079990236587,
"grad_norm": 0.27367883920669556,
"learning_rate": 2.3142382731423827e-05,
"loss": 1.8381233215332031,
"step": 4622
},
{
"epoch": 1.6123053855675855,
"grad_norm": 0.16515342891216278,
"learning_rate": 2.310087173100872e-05,
"loss": 1.8287198543548584,
"step": 4624
},
{
"epoch": 1.613002772111512,
"grad_norm": 0.2080453485250473,
"learning_rate": 2.3059360730593607e-05,
"loss": 1.9790563583374023,
"step": 4626
},
{
"epoch": 1.6137001586554387,
"grad_norm": 0.17708520591259003,
"learning_rate": 2.3017849730178497e-05,
"loss": 1.7650117874145508,
"step": 4628
},
{
"epoch": 1.6143975451993655,
"grad_norm": 0.19648979604244232,
"learning_rate": 2.297633872976339e-05,
"loss": 1.8148479461669922,
"step": 4630
},
{
"epoch": 1.6150949317432919,
"grad_norm": 0.21080146729946136,
"learning_rate": 2.293482772934828e-05,
"loss": 1.6323823928833008,
"step": 4632
},
{
"epoch": 1.6157923182872187,
"grad_norm": 0.19475789368152618,
"learning_rate": 2.2893316728933166e-05,
"loss": 1.8818817138671875,
"step": 4634
},
{
"epoch": 1.6164897048311453,
"grad_norm": 0.2431972622871399,
"learning_rate": 2.285180572851806e-05,
"loss": 1.8808767795562744,
"step": 4636
},
{
"epoch": 1.6171870913750719,
"grad_norm": 0.18643788993358612,
"learning_rate": 2.281029472810295e-05,
"loss": 2.0228090286254883,
"step": 4638
},
{
"epoch": 1.6178844779189987,
"grad_norm": 0.2250274419784546,
"learning_rate": 2.276878372768784e-05,
"loss": 1.9277596473693848,
"step": 4640
},
{
"epoch": 1.618581864462925,
"grad_norm": 0.18061144649982452,
"learning_rate": 2.272727272727273e-05,
"loss": 1.8646152019500732,
"step": 4642
},
{
"epoch": 1.6192792510068519,
"grad_norm": 0.20592698454856873,
"learning_rate": 2.268576172685762e-05,
"loss": 2.074492931365967,
"step": 4644
},
{
"epoch": 1.6199766375507785,
"grad_norm": 0.16534845530986786,
"learning_rate": 2.264425072644251e-05,
"loss": 1.836211919784546,
"step": 4646
},
{
"epoch": 1.620674024094705,
"grad_norm": 0.17768870294094086,
"learning_rate": 2.2602739726027396e-05,
"loss": 1.7464983463287354,
"step": 4648
},
{
"epoch": 1.6213714106386319,
"grad_norm": 0.1882151961326599,
"learning_rate": 2.256122872561229e-05,
"loss": 1.8208096027374268,
"step": 4650
},
{
"epoch": 1.6220687971825583,
"grad_norm": 0.20880545675754547,
"learning_rate": 2.251971772519718e-05,
"loss": 2.0152459144592285,
"step": 4652
},
{
"epoch": 1.622766183726485,
"grad_norm": 0.2227596938610077,
"learning_rate": 2.247820672478207e-05,
"loss": 1.7131097316741943,
"step": 4654
},
{
"epoch": 1.6234635702704117,
"grad_norm": 0.25815996527671814,
"learning_rate": 2.243669572436696e-05,
"loss": 1.9314322471618652,
"step": 4656
},
{
"epoch": 1.6241609568143383,
"grad_norm": 0.1772078424692154,
"learning_rate": 2.2395184723951848e-05,
"loss": 1.950426697731018,
"step": 4658
},
{
"epoch": 1.624858343358265,
"grad_norm": 0.28483498096466064,
"learning_rate": 2.2353673723536738e-05,
"loss": 1.7474815845489502,
"step": 4660
},
{
"epoch": 1.6255557299021914,
"grad_norm": 0.22062015533447266,
"learning_rate": 2.2312162723121628e-05,
"loss": 1.8343017101287842,
"step": 4662
},
{
"epoch": 1.6262531164461183,
"grad_norm": 0.22442767024040222,
"learning_rate": 2.2270651722706518e-05,
"loss": 2.063568115234375,
"step": 4664
},
{
"epoch": 1.6269505029900448,
"grad_norm": 0.16882704198360443,
"learning_rate": 2.2229140722291408e-05,
"loss": 1.734299898147583,
"step": 4666
},
{
"epoch": 1.6276478895339714,
"grad_norm": 0.32696422934532166,
"learning_rate": 2.2187629721876298e-05,
"loss": 1.8265082836151123,
"step": 4668
},
{
"epoch": 1.628345276077898,
"grad_norm": 0.24028272926807404,
"learning_rate": 2.2146118721461187e-05,
"loss": 1.8590327501296997,
"step": 4670
},
{
"epoch": 1.6290426626218246,
"grad_norm": 0.23825354874134064,
"learning_rate": 2.2104607721046077e-05,
"loss": 1.8434529304504395,
"step": 4672
},
{
"epoch": 1.6297400491657514,
"grad_norm": 0.1835913360118866,
"learning_rate": 2.2063096720630967e-05,
"loss": 2.043811082839966,
"step": 4674
},
{
"epoch": 1.630437435709678,
"grad_norm": 0.33232802152633667,
"learning_rate": 2.202158572021586e-05,
"loss": 1.423002004623413,
"step": 4676
},
{
"epoch": 1.6311348222536046,
"grad_norm": 0.2597466707229614,
"learning_rate": 2.1980074719800747e-05,
"loss": 1.5783658027648926,
"step": 4678
},
{
"epoch": 1.6318322087975312,
"grad_norm": 0.20854835212230682,
"learning_rate": 2.1938563719385637e-05,
"loss": 1.742499828338623,
"step": 4680
},
{
"epoch": 1.6325295953414578,
"grad_norm": 0.16686800122261047,
"learning_rate": 2.189705271897053e-05,
"loss": 1.9896137714385986,
"step": 4682
},
{
"epoch": 1.6332269818853846,
"grad_norm": 0.2372114509344101,
"learning_rate": 2.185554171855542e-05,
"loss": 1.9612038135528564,
"step": 4684
},
{
"epoch": 1.633924368429311,
"grad_norm": 0.2116309553384781,
"learning_rate": 2.1814030718140306e-05,
"loss": 2.0207488536834717,
"step": 4686
},
{
"epoch": 1.6346217549732378,
"grad_norm": 0.27684271335601807,
"learning_rate": 2.17725197177252e-05,
"loss": 1.9734842777252197,
"step": 4688
},
{
"epoch": 1.6353191415171644,
"grad_norm": 0.16166572272777557,
"learning_rate": 2.173100871731009e-05,
"loss": 1.9486138820648193,
"step": 4690
},
{
"epoch": 1.636016528061091,
"grad_norm": 0.21770936250686646,
"learning_rate": 2.1689497716894976e-05,
"loss": 1.8341920375823975,
"step": 4692
},
{
"epoch": 1.6367139146050178,
"grad_norm": 0.2177085131406784,
"learning_rate": 2.164798671647987e-05,
"loss": 1.7543977499008179,
"step": 4694
},
{
"epoch": 1.6374113011489442,
"grad_norm": 0.19301122426986694,
"learning_rate": 2.160647571606476e-05,
"loss": 1.9191703796386719,
"step": 4696
},
{
"epoch": 1.638108687692871,
"grad_norm": 0.30402037501335144,
"learning_rate": 2.156496471564965e-05,
"loss": 2.019595146179199,
"step": 4698
},
{
"epoch": 1.6388060742367976,
"grad_norm": 0.1700339913368225,
"learning_rate": 2.1523453715234536e-05,
"loss": 1.8973959684371948,
"step": 4700
},
{
"epoch": 1.6395034607807242,
"grad_norm": 0.2474794238805771,
"learning_rate": 2.148194271481943e-05,
"loss": 1.807422399520874,
"step": 4702
},
{
"epoch": 1.640200847324651,
"grad_norm": 0.18640539050102234,
"learning_rate": 2.144043171440432e-05,
"loss": 1.954378366470337,
"step": 4704
},
{
"epoch": 1.6408982338685774,
"grad_norm": 0.20491532981395721,
"learning_rate": 2.139892071398921e-05,
"loss": 2.0229756832122803,
"step": 4706
},
{
"epoch": 1.6415956204125042,
"grad_norm": 0.22327736020088196,
"learning_rate": 2.13574097135741e-05,
"loss": 1.999192476272583,
"step": 4708
},
{
"epoch": 1.6422930069564308,
"grad_norm": 0.2977581322193146,
"learning_rate": 2.1315898713158988e-05,
"loss": 1.7698019742965698,
"step": 4710
},
{
"epoch": 1.6429903935003574,
"grad_norm": 0.2158757448196411,
"learning_rate": 2.1274387712743878e-05,
"loss": 1.9217262268066406,
"step": 4712
},
{
"epoch": 1.6436877800442842,
"grad_norm": 0.33272016048431396,
"learning_rate": 2.1232876712328768e-05,
"loss": 1.839422583580017,
"step": 4714
},
{
"epoch": 1.6443851665882105,
"grad_norm": 0.1533578783273697,
"learning_rate": 2.1191365711913658e-05,
"loss": 1.7826305627822876,
"step": 4716
},
{
"epoch": 1.6450825531321374,
"grad_norm": 0.1968238353729248,
"learning_rate": 2.1149854711498548e-05,
"loss": 1.966490387916565,
"step": 4718
},
{
"epoch": 1.645779939676064,
"grad_norm": 0.14347350597381592,
"learning_rate": 2.1108343711083438e-05,
"loss": 1.8738987445831299,
"step": 4720
},
{
"epoch": 1.6464773262199905,
"grad_norm": 0.25829458236694336,
"learning_rate": 2.1066832710668328e-05,
"loss": 1.9003171920776367,
"step": 4722
},
{
"epoch": 1.6471747127639174,
"grad_norm": 0.2608783543109894,
"learning_rate": 2.1025321710253217e-05,
"loss": 1.831920862197876,
"step": 4724
},
{
"epoch": 1.6478720993078437,
"grad_norm": 0.2462533712387085,
"learning_rate": 2.0983810709838107e-05,
"loss": 1.9809796810150146,
"step": 4726
},
{
"epoch": 1.6485694858517705,
"grad_norm": 0.2654598653316498,
"learning_rate": 2.0942299709423e-05,
"loss": 1.7940154075622559,
"step": 4728
},
{
"epoch": 1.6492668723956971,
"grad_norm": 0.24551348388195038,
"learning_rate": 2.0900788709007887e-05,
"loss": 1.796177864074707,
"step": 4730
},
{
"epoch": 1.6499642589396237,
"grad_norm": 0.16254837810993195,
"learning_rate": 2.0859277708592777e-05,
"loss": 1.8673527240753174,
"step": 4732
},
{
"epoch": 1.6506616454835505,
"grad_norm": 0.3520248532295227,
"learning_rate": 2.081776670817767e-05,
"loss": 1.6647727489471436,
"step": 4734
},
{
"epoch": 1.651359032027477,
"grad_norm": 0.34911710023880005,
"learning_rate": 2.0776255707762557e-05,
"loss": 1.5715529918670654,
"step": 4736
},
{
"epoch": 1.6520564185714037,
"grad_norm": 0.2117566615343094,
"learning_rate": 2.0734744707347447e-05,
"loss": 1.7963566780090332,
"step": 4738
},
{
"epoch": 1.6527538051153303,
"grad_norm": 0.1970280557870865,
"learning_rate": 2.069323370693234e-05,
"loss": 1.5644910335540771,
"step": 4740
},
{
"epoch": 1.653451191659257,
"grad_norm": 0.22263410687446594,
"learning_rate": 2.065172270651723e-05,
"loss": 1.914230465888977,
"step": 4742
},
{
"epoch": 1.6541485782031837,
"grad_norm": 0.15990054607391357,
"learning_rate": 2.0610211706102116e-05,
"loss": 1.8632224798202515,
"step": 4744
},
{
"epoch": 1.65484596474711,
"grad_norm": 0.26392489671707153,
"learning_rate": 2.056870070568701e-05,
"loss": 1.6430706977844238,
"step": 4746
},
{
"epoch": 1.655543351291037,
"grad_norm": 0.20190612971782684,
"learning_rate": 2.05271897052719e-05,
"loss": 1.8972723484039307,
"step": 4748
},
{
"epoch": 1.6562407378349635,
"grad_norm": 0.23242701590061188,
"learning_rate": 2.048567870485679e-05,
"loss": 1.948430061340332,
"step": 4750
},
{
"epoch": 1.65693812437889,
"grad_norm": 0.1872187703847885,
"learning_rate": 2.044416770444168e-05,
"loss": 1.943682074546814,
"step": 4752
},
{
"epoch": 1.6576355109228167,
"grad_norm": 0.3118439316749573,
"learning_rate": 2.040265670402657e-05,
"loss": 1.8686903715133667,
"step": 4754
},
{
"epoch": 1.6583328974667433,
"grad_norm": 0.24208366870880127,
"learning_rate": 2.036114570361146e-05,
"loss": 1.846539855003357,
"step": 4756
},
{
"epoch": 1.65903028401067,
"grad_norm": 0.166546031832695,
"learning_rate": 2.0319634703196345e-05,
"loss": 1.9614691734313965,
"step": 4758
},
{
"epoch": 1.6597276705545967,
"grad_norm": 0.26857540011405945,
"learning_rate": 2.027812370278124e-05,
"loss": 1.7173560857772827,
"step": 4760
},
{
"epoch": 1.6604250570985233,
"grad_norm": 0.18927040696144104,
"learning_rate": 2.023661270236613e-05,
"loss": 1.968058705329895,
"step": 4762
},
{
"epoch": 1.6611224436424499,
"grad_norm": 0.20829182863235474,
"learning_rate": 2.0195101701951018e-05,
"loss": 2.138188362121582,
"step": 4764
},
{
"epoch": 1.6618198301863765,
"grad_norm": 0.26986241340637207,
"learning_rate": 2.0153590701535908e-05,
"loss": 1.6226156949996948,
"step": 4766
},
{
"epoch": 1.6625172167303033,
"grad_norm": 0.29422932863235474,
"learning_rate": 2.0112079701120798e-05,
"loss": 1.849477767944336,
"step": 4768
},
{
"epoch": 1.6632146032742299,
"grad_norm": 0.19077733159065247,
"learning_rate": 2.0070568700705688e-05,
"loss": 1.795039415359497,
"step": 4770
},
{
"epoch": 1.6639119898181565,
"grad_norm": 0.2797706425189972,
"learning_rate": 2.0029057700290578e-05,
"loss": 1.8383277654647827,
"step": 4772
},
{
"epoch": 1.664609376362083,
"grad_norm": 0.3744497001171112,
"learning_rate": 1.9987546699875468e-05,
"loss": 1.886944055557251,
"step": 4774
},
{
"epoch": 1.6653067629060097,
"grad_norm": 0.24865016341209412,
"learning_rate": 1.9946035699460357e-05,
"loss": 1.7954492568969727,
"step": 4776
},
{
"epoch": 1.6660041494499365,
"grad_norm": 0.3195587396621704,
"learning_rate": 1.9904524699045247e-05,
"loss": 1.9134800434112549,
"step": 4778
},
{
"epoch": 1.6667015359938628,
"grad_norm": 0.1610836535692215,
"learning_rate": 1.9863013698630137e-05,
"loss": 2.004275321960449,
"step": 4780
},
{
"epoch": 1.6673989225377897,
"grad_norm": 0.23241788148880005,
"learning_rate": 1.9821502698215027e-05,
"loss": 1.7410321235656738,
"step": 4782
},
{
"epoch": 1.6680963090817162,
"grad_norm": 0.28802603483200073,
"learning_rate": 1.9779991697799917e-05,
"loss": 1.9949110746383667,
"step": 4784
},
{
"epoch": 1.6687936956256428,
"grad_norm": 0.25853458046913147,
"learning_rate": 1.973848069738481e-05,
"loss": 1.591475009918213,
"step": 4786
},
{
"epoch": 1.6694910821695697,
"grad_norm": 0.2779485583305359,
"learning_rate": 1.9696969696969697e-05,
"loss": 1.800331950187683,
"step": 4788
},
{
"epoch": 1.670188468713496,
"grad_norm": 0.2355402410030365,
"learning_rate": 1.9655458696554587e-05,
"loss": 1.7988674640655518,
"step": 4790
},
{
"epoch": 1.6708858552574228,
"grad_norm": 0.250661700963974,
"learning_rate": 1.961394769613948e-05,
"loss": 1.9616501331329346,
"step": 4792
},
{
"epoch": 1.6715832418013494,
"grad_norm": 0.1739974170923233,
"learning_rate": 1.9572436695724366e-05,
"loss": 1.7613005638122559,
"step": 4794
},
{
"epoch": 1.672280628345276,
"grad_norm": 0.1977234184741974,
"learning_rate": 1.9530925695309256e-05,
"loss": 2.229280471801758,
"step": 4796
},
{
"epoch": 1.6729780148892028,
"grad_norm": 0.3321165442466736,
"learning_rate": 1.948941469489415e-05,
"loss": 1.6390643119812012,
"step": 4798
},
{
"epoch": 1.6736754014331292,
"grad_norm": 0.2189665585756302,
"learning_rate": 1.944790369447904e-05,
"loss": 1.80131995677948,
"step": 4800
},
{
"epoch": 1.674372787977056,
"grad_norm": 0.24850307404994965,
"learning_rate": 1.9406392694063926e-05,
"loss": 1.7933447360992432,
"step": 4802
},
{
"epoch": 1.6750701745209826,
"grad_norm": 0.16121813654899597,
"learning_rate": 1.936488169364882e-05,
"loss": 2.0353100299835205,
"step": 4804
},
{
"epoch": 1.6757675610649092,
"grad_norm": 0.1946568489074707,
"learning_rate": 1.932337069323371e-05,
"loss": 1.897767186164856,
"step": 4806
},
{
"epoch": 1.676464947608836,
"grad_norm": 0.18960946798324585,
"learning_rate": 1.92818596928186e-05,
"loss": 1.7468819618225098,
"step": 4808
},
{
"epoch": 1.6771623341527624,
"grad_norm": 0.21749210357666016,
"learning_rate": 1.924034869240349e-05,
"loss": 2.0478451251983643,
"step": 4810
},
{
"epoch": 1.6778597206966892,
"grad_norm": 0.25406748056411743,
"learning_rate": 1.919883769198838e-05,
"loss": 1.8166673183441162,
"step": 4812
},
{
"epoch": 1.6785571072406158,
"grad_norm": 0.2774946689605713,
"learning_rate": 1.915732669157327e-05,
"loss": 1.8128101825714111,
"step": 4814
},
{
"epoch": 1.6792544937845424,
"grad_norm": 0.20086419582366943,
"learning_rate": 1.9115815691158155e-05,
"loss": 1.9278628826141357,
"step": 4816
},
{
"epoch": 1.6799518803284692,
"grad_norm": 0.2466057986021042,
"learning_rate": 1.9074304690743048e-05,
"loss": 1.779740571975708,
"step": 4818
},
{
"epoch": 1.6806492668723956,
"grad_norm": 0.23086068034172058,
"learning_rate": 1.9032793690327938e-05,
"loss": 1.7101213932037354,
"step": 4820
},
{
"epoch": 1.6813466534163224,
"grad_norm": 0.25265970826148987,
"learning_rate": 1.8991282689912828e-05,
"loss": 1.8596704006195068,
"step": 4822
},
{
"epoch": 1.682044039960249,
"grad_norm": 0.27982133626937866,
"learning_rate": 1.8949771689497718e-05,
"loss": 1.7157399654388428,
"step": 4824
},
{
"epoch": 1.6827414265041756,
"grad_norm": 0.19305096566677094,
"learning_rate": 1.8908260689082608e-05,
"loss": 1.8414602279663086,
"step": 4826
},
{
"epoch": 1.6834388130481024,
"grad_norm": 0.20308367908000946,
"learning_rate": 1.8866749688667497e-05,
"loss": 1.90911865234375,
"step": 4828
},
{
"epoch": 1.6841361995920288,
"grad_norm": 0.20568597316741943,
"learning_rate": 1.8825238688252387e-05,
"loss": 1.867128849029541,
"step": 4830
},
{
"epoch": 1.6848335861359556,
"grad_norm": 0.2156611531972885,
"learning_rate": 1.8783727687837277e-05,
"loss": 1.8440172672271729,
"step": 4832
},
{
"epoch": 1.6855309726798822,
"grad_norm": 0.20552696287631989,
"learning_rate": 1.8742216687422167e-05,
"loss": 1.835031509399414,
"step": 4834
},
{
"epoch": 1.6862283592238088,
"grad_norm": 0.19234976172447205,
"learning_rate": 1.8700705687007057e-05,
"loss": 1.8685534000396729,
"step": 4836
},
{
"epoch": 1.6869257457677356,
"grad_norm": 0.1943400353193283,
"learning_rate": 1.8659194686591947e-05,
"loss": 1.8733526468276978,
"step": 4838
},
{
"epoch": 1.687623132311662,
"grad_norm": 0.23722796142101288,
"learning_rate": 1.8617683686176837e-05,
"loss": 1.9736058712005615,
"step": 4840
},
{
"epoch": 1.6883205188555888,
"grad_norm": 0.2323332130908966,
"learning_rate": 1.8576172685761727e-05,
"loss": 1.9442733526229858,
"step": 4842
},
{
"epoch": 1.6890179053995154,
"grad_norm": 0.3275192975997925,
"learning_rate": 1.853466168534662e-05,
"loss": 1.8706843852996826,
"step": 4844
},
{
"epoch": 1.689715291943442,
"grad_norm": 0.21366503834724426,
"learning_rate": 1.8493150684931506e-05,
"loss": 1.7669522762298584,
"step": 4846
},
{
"epoch": 1.6904126784873685,
"grad_norm": 0.21341843903064728,
"learning_rate": 1.8451639684516396e-05,
"loss": 1.9052839279174805,
"step": 4848
},
{
"epoch": 1.6911100650312951,
"grad_norm": 0.20396091043949127,
"learning_rate": 1.841012868410129e-05,
"loss": 2.0447299480438232,
"step": 4850
},
{
"epoch": 1.691807451575222,
"grad_norm": 0.25687476992607117,
"learning_rate": 1.836861768368618e-05,
"loss": 1.817075490951538,
"step": 4852
},
{
"epoch": 1.6925048381191485,
"grad_norm": 0.21825337409973145,
"learning_rate": 1.8327106683271066e-05,
"loss": 1.9545912742614746,
"step": 4854
},
{
"epoch": 1.6932022246630751,
"grad_norm": 0.23390546441078186,
"learning_rate": 1.828559568285596e-05,
"loss": 1.8302881717681885,
"step": 4856
},
{
"epoch": 1.6938996112070017,
"grad_norm": 0.21338598430156708,
"learning_rate": 1.824408468244085e-05,
"loss": 1.9971953630447388,
"step": 4858
},
{
"epoch": 1.6945969977509283,
"grad_norm": 0.2040199190378189,
"learning_rate": 1.8202573682025735e-05,
"loss": 1.872685194015503,
"step": 4860
},
{
"epoch": 1.6952943842948551,
"grad_norm": 0.21568608283996582,
"learning_rate": 1.816106268161063e-05,
"loss": 1.9188270568847656,
"step": 4862
},
{
"epoch": 1.6959917708387817,
"grad_norm": 0.2995471954345703,
"learning_rate": 1.811955168119552e-05,
"loss": 1.8098963499069214,
"step": 4864
},
{
"epoch": 1.6966891573827083,
"grad_norm": 0.2966385781764984,
"learning_rate": 1.807804068078041e-05,
"loss": 1.6882479190826416,
"step": 4866
},
{
"epoch": 1.697386543926635,
"grad_norm": 0.1585799902677536,
"learning_rate": 1.80365296803653e-05,
"loss": 1.69493567943573,
"step": 4868
},
{
"epoch": 1.6980839304705615,
"grad_norm": 0.21754221618175507,
"learning_rate": 1.7995018679950188e-05,
"loss": 1.9639341831207275,
"step": 4870
},
{
"epoch": 1.6987813170144883,
"grad_norm": 0.302188903093338,
"learning_rate": 1.7953507679535078e-05,
"loss": 1.8428955078125,
"step": 4872
},
{
"epoch": 1.6994787035584147,
"grad_norm": 0.25416743755340576,
"learning_rate": 1.7911996679119968e-05,
"loss": 1.735520601272583,
"step": 4874
},
{
"epoch": 1.7001760901023415,
"grad_norm": 0.1922425776720047,
"learning_rate": 1.7870485678704858e-05,
"loss": 1.8143341541290283,
"step": 4876
},
{
"epoch": 1.700873476646268,
"grad_norm": 0.21087424457073212,
"learning_rate": 1.7828974678289748e-05,
"loss": 1.6546618938446045,
"step": 4878
},
{
"epoch": 1.7015708631901947,
"grad_norm": 0.18634897470474243,
"learning_rate": 1.7787463677874638e-05,
"loss": 1.892411231994629,
"step": 4880
},
{
"epoch": 1.7022682497341215,
"grad_norm": 0.22964993119239807,
"learning_rate": 1.7745952677459527e-05,
"loss": 1.7684986591339111,
"step": 4882
},
{
"epoch": 1.7029656362780479,
"grad_norm": 0.16608180105686188,
"learning_rate": 1.7704441677044417e-05,
"loss": 1.7425634860992432,
"step": 4884
},
{
"epoch": 1.7036630228219747,
"grad_norm": 0.1754094660282135,
"learning_rate": 1.7662930676629307e-05,
"loss": 1.9515323638916016,
"step": 4886
},
{
"epoch": 1.7043604093659013,
"grad_norm": 0.2333822101354599,
"learning_rate": 1.7621419676214197e-05,
"loss": 1.8986519575119019,
"step": 4888
},
{
"epoch": 1.7050577959098279,
"grad_norm": 0.20273318886756897,
"learning_rate": 1.7579908675799087e-05,
"loss": 2.0065834522247314,
"step": 4890
},
{
"epoch": 1.7057551824537547,
"grad_norm": 0.20620377361774445,
"learning_rate": 1.7538397675383977e-05,
"loss": 2.0093650817871094,
"step": 4892
},
{
"epoch": 1.706452568997681,
"grad_norm": 0.24058882892131805,
"learning_rate": 1.7496886674968867e-05,
"loss": 1.8212504386901855,
"step": 4894
},
{
"epoch": 1.7071499555416079,
"grad_norm": 0.15788443386554718,
"learning_rate": 1.745537567455376e-05,
"loss": 1.874027967453003,
"step": 4896
},
{
"epoch": 1.7078473420855345,
"grad_norm": 0.2806450426578522,
"learning_rate": 1.7413864674138646e-05,
"loss": 1.8924381732940674,
"step": 4898
},
{
"epoch": 1.708544728629461,
"grad_norm": 0.2721128761768341,
"learning_rate": 1.7372353673723536e-05,
"loss": 1.6524205207824707,
"step": 4900
},
{
"epoch": 1.7092421151733879,
"grad_norm": 0.15610946714878082,
"learning_rate": 1.733084267330843e-05,
"loss": 1.7538938522338867,
"step": 4902
},
{
"epoch": 1.7099395017173142,
"grad_norm": 0.24323315918445587,
"learning_rate": 1.7289331672893316e-05,
"loss": 1.9989988803863525,
"step": 4904
},
{
"epoch": 1.710636888261241,
"grad_norm": 0.22363847494125366,
"learning_rate": 1.7247820672478206e-05,
"loss": 1.8541325330734253,
"step": 4906
},
{
"epoch": 1.7113342748051676,
"grad_norm": 0.21976357698440552,
"learning_rate": 1.72063096720631e-05,
"loss": 1.7877877950668335,
"step": 4908
},
{
"epoch": 1.7120316613490942,
"grad_norm": 0.24294093251228333,
"learning_rate": 1.716479867164799e-05,
"loss": 1.7813832759857178,
"step": 4910
},
{
"epoch": 1.712729047893021,
"grad_norm": 0.3119526505470276,
"learning_rate": 1.7123287671232875e-05,
"loss": 1.5923690795898438,
"step": 4912
},
{
"epoch": 1.7134264344369474,
"grad_norm": 0.18126317858695984,
"learning_rate": 1.708177667081777e-05,
"loss": 1.9581480026245117,
"step": 4914
},
{
"epoch": 1.7141238209808742,
"grad_norm": 0.1903887689113617,
"learning_rate": 1.704026567040266e-05,
"loss": 1.8188350200653076,
"step": 4916
},
{
"epoch": 1.7148212075248008,
"grad_norm": 0.27518969774246216,
"learning_rate": 1.699875466998755e-05,
"loss": 1.9018728733062744,
"step": 4918
},
{
"epoch": 1.7155185940687274,
"grad_norm": 0.27955564856529236,
"learning_rate": 1.695724366957244e-05,
"loss": 1.852898359298706,
"step": 4920
},
{
"epoch": 1.7162159806126542,
"grad_norm": 0.2705944776535034,
"learning_rate": 1.6915732669157328e-05,
"loss": 2.041968822479248,
"step": 4922
},
{
"epoch": 1.7169133671565806,
"grad_norm": 0.1837584525346756,
"learning_rate": 1.6874221668742218e-05,
"loss": 1.9442403316497803,
"step": 4924
},
{
"epoch": 1.7176107537005074,
"grad_norm": 0.1979951709508896,
"learning_rate": 1.6832710668327108e-05,
"loss": 2.0591835975646973,
"step": 4926
},
{
"epoch": 1.718308140244434,
"grad_norm": 0.19860365986824036,
"learning_rate": 1.6791199667911998e-05,
"loss": 2.0726473331451416,
"step": 4928
},
{
"epoch": 1.7190055267883606,
"grad_norm": 0.23286722600460052,
"learning_rate": 1.6749688667496888e-05,
"loss": 1.8184525966644287,
"step": 4930
},
{
"epoch": 1.7197029133322874,
"grad_norm": 0.20031099021434784,
"learning_rate": 1.6708177667081778e-05,
"loss": 1.8004885911941528,
"step": 4932
},
{
"epoch": 1.7204002998762138,
"grad_norm": 0.24718044698238373,
"learning_rate": 1.6666666666666667e-05,
"loss": 1.4946333169937134,
"step": 4934
},
{
"epoch": 1.7210976864201406,
"grad_norm": 0.1813986599445343,
"learning_rate": 1.6625155666251557e-05,
"loss": 1.8198713064193726,
"step": 4936
},
{
"epoch": 1.7217950729640672,
"grad_norm": 0.1980566829442978,
"learning_rate": 1.6583644665836447e-05,
"loss": 1.9180089235305786,
"step": 4938
},
{
"epoch": 1.7224924595079938,
"grad_norm": 0.16343583166599274,
"learning_rate": 1.654213366542134e-05,
"loss": 1.9447441101074219,
"step": 4940
},
{
"epoch": 1.7231898460519204,
"grad_norm": 0.22005994617938995,
"learning_rate": 1.6500622665006227e-05,
"loss": 2.0293641090393066,
"step": 4942
},
{
"epoch": 1.723887232595847,
"grad_norm": 0.21437813341617584,
"learning_rate": 1.6459111664591117e-05,
"loss": 2.0307180881500244,
"step": 4944
},
{
"epoch": 1.7245846191397738,
"grad_norm": 0.19174791872501373,
"learning_rate": 1.6417600664176007e-05,
"loss": 1.8691425323486328,
"step": 4946
},
{
"epoch": 1.7252820056837004,
"grad_norm": 0.23107494413852692,
"learning_rate": 1.6376089663760897e-05,
"loss": 1.7415695190429688,
"step": 4948
},
{
"epoch": 1.725979392227627,
"grad_norm": 0.16872994601726532,
"learning_rate": 1.6334578663345786e-05,
"loss": 1.952255129814148,
"step": 4950
},
{
"epoch": 1.7266767787715536,
"grad_norm": 0.21751785278320312,
"learning_rate": 1.6293067662930676e-05,
"loss": 1.8491461277008057,
"step": 4952
},
{
"epoch": 1.7273741653154802,
"grad_norm": 0.17315080761909485,
"learning_rate": 1.625155666251557e-05,
"loss": 1.978123426437378,
"step": 4954
},
{
"epoch": 1.728071551859407,
"grad_norm": 0.2615652084350586,
"learning_rate": 1.6210045662100456e-05,
"loss": 1.8091578483581543,
"step": 4956
},
{
"epoch": 1.7287689384033333,
"grad_norm": 0.24388821423053741,
"learning_rate": 1.6168534661685346e-05,
"loss": 1.6740353107452393,
"step": 4958
},
{
"epoch": 1.7294663249472602,
"grad_norm": 0.21198374032974243,
"learning_rate": 1.612702366127024e-05,
"loss": 1.9332044124603271,
"step": 4960
},
{
"epoch": 1.7301637114911868,
"grad_norm": 0.19904807209968567,
"learning_rate": 1.608551266085513e-05,
"loss": 1.9341310262680054,
"step": 4962
},
{
"epoch": 1.7308610980351133,
"grad_norm": 0.2688181400299072,
"learning_rate": 1.6044001660440016e-05,
"loss": 1.7314822673797607,
"step": 4964
},
{
"epoch": 1.7315584845790402,
"grad_norm": 0.22682805359363556,
"learning_rate": 1.600249066002491e-05,
"loss": 1.7970565557479858,
"step": 4966
},
{
"epoch": 1.7322558711229665,
"grad_norm": 0.22815777361392975,
"learning_rate": 1.59609796596098e-05,
"loss": 1.8633747100830078,
"step": 4968
},
{
"epoch": 1.7329532576668933,
"grad_norm": 0.19783873856067657,
"learning_rate": 1.5919468659194685e-05,
"loss": 1.6591200828552246,
"step": 4970
},
{
"epoch": 1.73365064421082,
"grad_norm": 0.2341974824666977,
"learning_rate": 1.587795765877958e-05,
"loss": 1.9787203073501587,
"step": 4972
},
{
"epoch": 1.7343480307547465,
"grad_norm": 0.16734273731708527,
"learning_rate": 1.5836446658364468e-05,
"loss": 2.0457887649536133,
"step": 4974
},
{
"epoch": 1.7350454172986733,
"grad_norm": 0.17628929018974304,
"learning_rate": 1.5794935657949358e-05,
"loss": 1.8850793838500977,
"step": 4976
},
{
"epoch": 1.7357428038425997,
"grad_norm": 0.24273666739463806,
"learning_rate": 1.5753424657534248e-05,
"loss": 1.8593032360076904,
"step": 4978
},
{
"epoch": 1.7364401903865265,
"grad_norm": 0.2080320417881012,
"learning_rate": 1.5711913657119138e-05,
"loss": 1.899301528930664,
"step": 4980
},
{
"epoch": 1.7371375769304531,
"grad_norm": 0.23769371211528778,
"learning_rate": 1.5670402656704028e-05,
"loss": 1.7426323890686035,
"step": 4982
},
{
"epoch": 1.7378349634743797,
"grad_norm": 0.2175651341676712,
"learning_rate": 1.5628891656288914e-05,
"loss": 2.0780487060546875,
"step": 4984
},
{
"epoch": 1.7385323500183065,
"grad_norm": 0.2556368112564087,
"learning_rate": 1.5587380655873808e-05,
"loss": 1.7094303369522095,
"step": 4986
},
{
"epoch": 1.739229736562233,
"grad_norm": 0.17907623946666718,
"learning_rate": 1.5545869655458697e-05,
"loss": 2.028717041015625,
"step": 4988
},
{
"epoch": 1.7399271231061597,
"grad_norm": 0.21630629897117615,
"learning_rate": 1.5504358655043587e-05,
"loss": 2.0060858726501465,
"step": 4990
},
{
"epoch": 1.7406245096500863,
"grad_norm": 0.1869284212589264,
"learning_rate": 1.5462847654628477e-05,
"loss": 1.8111498355865479,
"step": 4992
},
{
"epoch": 1.741321896194013,
"grad_norm": 0.2576078176498413,
"learning_rate": 1.5421336654213367e-05,
"loss": 1.8613758087158203,
"step": 4994
},
{
"epoch": 1.7420192827379397,
"grad_norm": 0.190448597073555,
"learning_rate": 1.5379825653798257e-05,
"loss": 1.921868085861206,
"step": 4996
},
{
"epoch": 1.742716669281866,
"grad_norm": 0.18180961906909943,
"learning_rate": 1.533831465338315e-05,
"loss": 1.9190137386322021,
"step": 4998
},
{
"epoch": 1.743414055825793,
"grad_norm": 0.25628480315208435,
"learning_rate": 1.5296803652968037e-05,
"loss": 2.04146146774292,
"step": 5000
},
{
"epoch": 1.7441114423697195,
"grad_norm": 0.17649434506893158,
"learning_rate": 1.5255292652552926e-05,
"loss": 2.016183853149414,
"step": 5002
},
{
"epoch": 1.744808828913646,
"grad_norm": 0.22032147645950317,
"learning_rate": 1.5213781652137818e-05,
"loss": 1.842036485671997,
"step": 5004
},
{
"epoch": 1.745506215457573,
"grad_norm": 0.22300808131694794,
"learning_rate": 1.5172270651722706e-05,
"loss": 1.7307430505752563,
"step": 5006
},
{
"epoch": 1.7462036020014993,
"grad_norm": 0.14669063687324524,
"learning_rate": 1.5130759651307596e-05,
"loss": 1.892002820968628,
"step": 5008
},
{
"epoch": 1.746900988545426,
"grad_norm": 0.1940767765045166,
"learning_rate": 1.5089248650892488e-05,
"loss": 1.7894619703292847,
"step": 5010
},
{
"epoch": 1.7475983750893527,
"grad_norm": 0.20655557513237,
"learning_rate": 1.5047737650477378e-05,
"loss": 1.6511226892471313,
"step": 5012
},
{
"epoch": 1.7482957616332793,
"grad_norm": 0.2275824099779129,
"learning_rate": 1.5006226650062266e-05,
"loss": 1.7030476331710815,
"step": 5014
},
{
"epoch": 1.748993148177206,
"grad_norm": 0.19695813953876495,
"learning_rate": 1.4964715649647157e-05,
"loss": 1.9162763357162476,
"step": 5016
},
{
"epoch": 1.7496905347211325,
"grad_norm": 0.1983014941215515,
"learning_rate": 1.4923204649232047e-05,
"loss": 1.8794071674346924,
"step": 5018
},
{
"epoch": 1.7503879212650593,
"grad_norm": 0.2710726857185364,
"learning_rate": 1.4881693648816939e-05,
"loss": 1.5816371440887451,
"step": 5020
},
{
"epoch": 1.7510853078089859,
"grad_norm": 0.21617639064788818,
"learning_rate": 1.4840182648401827e-05,
"loss": 1.8407632112503052,
"step": 5022
},
{
"epoch": 1.7517826943529125,
"grad_norm": 0.17437605559825897,
"learning_rate": 1.4798671647986717e-05,
"loss": 2.030404567718506,
"step": 5024
},
{
"epoch": 1.752480080896839,
"grad_norm": 0.18162742257118225,
"learning_rate": 1.4757160647571608e-05,
"loss": 1.889998197555542,
"step": 5026
},
{
"epoch": 1.7531774674407656,
"grad_norm": 0.30704429745674133,
"learning_rate": 1.4715649647156497e-05,
"loss": 1.7727937698364258,
"step": 5028
},
{
"epoch": 1.7538748539846925,
"grad_norm": 0.26485392451286316,
"learning_rate": 1.4674138646741386e-05,
"loss": 1.994034767150879,
"step": 5030
},
{
"epoch": 1.754572240528619,
"grad_norm": 0.2432965487241745,
"learning_rate": 1.4632627646326278e-05,
"loss": 1.752293586730957,
"step": 5032
},
{
"epoch": 1.7552696270725456,
"grad_norm": 0.19498766958713531,
"learning_rate": 1.4591116645911168e-05,
"loss": 1.9810562133789062,
"step": 5034
},
{
"epoch": 1.7559670136164722,
"grad_norm": 0.21565671265125275,
"learning_rate": 1.4549605645496056e-05,
"loss": 1.656402826309204,
"step": 5036
},
{
"epoch": 1.7566644001603988,
"grad_norm": 0.18841609358787537,
"learning_rate": 1.4508094645080948e-05,
"loss": 1.9723447561264038,
"step": 5038
},
{
"epoch": 1.7573617867043256,
"grad_norm": 0.20149941742420197,
"learning_rate": 1.4466583644665837e-05,
"loss": 1.699404239654541,
"step": 5040
},
{
"epoch": 1.7580591732482522,
"grad_norm": 0.2105809599161148,
"learning_rate": 1.4425072644250729e-05,
"loss": 1.9500550031661987,
"step": 5042
},
{
"epoch": 1.7587565597921788,
"grad_norm": 0.27240362763404846,
"learning_rate": 1.4383561643835617e-05,
"loss": 1.9278383255004883,
"step": 5044
},
{
"epoch": 1.7594539463361054,
"grad_norm": 0.19991464912891388,
"learning_rate": 1.4342050643420507e-05,
"loss": 1.8733749389648438,
"step": 5046
},
{
"epoch": 1.760151332880032,
"grad_norm": 0.3012697696685791,
"learning_rate": 1.4300539643005399e-05,
"loss": 1.8668156862258911,
"step": 5048
},
{
"epoch": 1.7608487194239588,
"grad_norm": 0.15615016222000122,
"learning_rate": 1.4259028642590285e-05,
"loss": 1.7432570457458496,
"step": 5050
},
{
"epoch": 1.7615461059678852,
"grad_norm": 0.21919383108615875,
"learning_rate": 1.4217517642175177e-05,
"loss": 1.748333215713501,
"step": 5052
},
{
"epoch": 1.762243492511812,
"grad_norm": 0.2517048120498657,
"learning_rate": 1.4176006641760067e-05,
"loss": 1.631244421005249,
"step": 5054
},
{
"epoch": 1.7629408790557386,
"grad_norm": 0.1723175048828125,
"learning_rate": 1.4134495641344958e-05,
"loss": 2.193490982055664,
"step": 5056
},
{
"epoch": 1.7636382655996652,
"grad_norm": 0.2375682294368744,
"learning_rate": 1.4092984640929846e-05,
"loss": 1.8086185455322266,
"step": 5058
},
{
"epoch": 1.764335652143592,
"grad_norm": 0.22081787884235382,
"learning_rate": 1.4051473640514736e-05,
"loss": 1.624267339706421,
"step": 5060
},
{
"epoch": 1.7650330386875184,
"grad_norm": 0.22836139798164368,
"learning_rate": 1.4009962640099628e-05,
"loss": 1.81376314163208,
"step": 5062
},
{
"epoch": 1.7657304252314452,
"grad_norm": 0.2079772800207138,
"learning_rate": 1.3968451639684518e-05,
"loss": 1.9828565120697021,
"step": 5064
},
{
"epoch": 1.7664278117753718,
"grad_norm": 0.2522355616092682,
"learning_rate": 1.3926940639269406e-05,
"loss": 1.907923698425293,
"step": 5066
},
{
"epoch": 1.7671251983192984,
"grad_norm": 0.26948314905166626,
"learning_rate": 1.3885429638854297e-05,
"loss": 1.6880152225494385,
"step": 5068
},
{
"epoch": 1.7678225848632252,
"grad_norm": 0.23745985329151154,
"learning_rate": 1.3843918638439187e-05,
"loss": 1.9411509037017822,
"step": 5070
},
{
"epoch": 1.7685199714071516,
"grad_norm": 0.21113702654838562,
"learning_rate": 1.3802407638024075e-05,
"loss": 1.9977704286575317,
"step": 5072
},
{
"epoch": 1.7692173579510784,
"grad_norm": 0.19825154542922974,
"learning_rate": 1.3760896637608967e-05,
"loss": 1.8850080966949463,
"step": 5074
},
{
"epoch": 1.769914744495005,
"grad_norm": 0.1835787296295166,
"learning_rate": 1.3719385637193857e-05,
"loss": 1.9037789106369019,
"step": 5076
},
{
"epoch": 1.7706121310389316,
"grad_norm": 0.26366475224494934,
"learning_rate": 1.3677874636778748e-05,
"loss": 1.7554395198822021,
"step": 5078
},
{
"epoch": 1.7713095175828584,
"grad_norm": 0.30773717164993286,
"learning_rate": 1.3636363636363637e-05,
"loss": 1.742945671081543,
"step": 5080
},
{
"epoch": 1.7720069041267847,
"grad_norm": 0.18697725236415863,
"learning_rate": 1.3594852635948526e-05,
"loss": 1.9001011848449707,
"step": 5082
},
{
"epoch": 1.7727042906707116,
"grad_norm": 0.24588505923748016,
"learning_rate": 1.3553341635533418e-05,
"loss": 1.6992133855819702,
"step": 5084
},
{
"epoch": 1.7734016772146382,
"grad_norm": 0.17033891379833221,
"learning_rate": 1.3511830635118308e-05,
"loss": 1.85475492477417,
"step": 5086
},
{
"epoch": 1.7740990637585647,
"grad_norm": 0.26573145389556885,
"learning_rate": 1.3470319634703196e-05,
"loss": 1.7976782321929932,
"step": 5088
},
{
"epoch": 1.7747964503024916,
"grad_norm": 0.18720924854278564,
"learning_rate": 1.3428808634288088e-05,
"loss": 1.8966387510299683,
"step": 5090
},
{
"epoch": 1.775493836846418,
"grad_norm": 0.2379140406847,
"learning_rate": 1.3387297633872977e-05,
"loss": 2.0105175971984863,
"step": 5092
},
{
"epoch": 1.7761912233903447,
"grad_norm": 0.23965798318386078,
"learning_rate": 1.3345786633457866e-05,
"loss": 1.7327382564544678,
"step": 5094
},
{
"epoch": 1.7768886099342713,
"grad_norm": 0.25432243943214417,
"learning_rate": 1.3304275633042757e-05,
"loss": 1.727048635482788,
"step": 5096
},
{
"epoch": 1.777585996478198,
"grad_norm": 0.23283375799655914,
"learning_rate": 1.3262764632627647e-05,
"loss": 2.0329108238220215,
"step": 5098
},
{
"epoch": 1.7782833830221247,
"grad_norm": 0.2430335134267807,
"learning_rate": 1.3221253632212539e-05,
"loss": 1.801249623298645,
"step": 5100
},
{
"epoch": 1.7789807695660511,
"grad_norm": 0.1487344354391098,
"learning_rate": 1.3179742631797425e-05,
"loss": 1.9199151992797852,
"step": 5102
},
{
"epoch": 1.779678156109978,
"grad_norm": 0.21184313297271729,
"learning_rate": 1.3138231631382317e-05,
"loss": 1.9471790790557861,
"step": 5104
},
{
"epoch": 1.7803755426539045,
"grad_norm": 0.1828693300485611,
"learning_rate": 1.3096720630967208e-05,
"loss": 2.0621299743652344,
"step": 5106
},
{
"epoch": 1.7810729291978311,
"grad_norm": 0.24108515679836273,
"learning_rate": 1.3055209630552098e-05,
"loss": 2.010344982147217,
"step": 5108
},
{
"epoch": 1.781770315741758,
"grad_norm": 0.27066123485565186,
"learning_rate": 1.3013698630136986e-05,
"loss": 1.695648431777954,
"step": 5110
},
{
"epoch": 1.7824677022856843,
"grad_norm": 0.2320084273815155,
"learning_rate": 1.2972187629721876e-05,
"loss": 2.066194772720337,
"step": 5112
},
{
"epoch": 1.7831650888296111,
"grad_norm": 0.20274823904037476,
"learning_rate": 1.2930676629306768e-05,
"loss": 1.9236104488372803,
"step": 5114
},
{
"epoch": 1.7838624753735377,
"grad_norm": 0.20278611779212952,
"learning_rate": 1.2889165628891656e-05,
"loss": 1.9049060344696045,
"step": 5116
},
{
"epoch": 1.7845598619174643,
"grad_norm": 0.27501341700553894,
"learning_rate": 1.2847654628476546e-05,
"loss": 1.7823734283447266,
"step": 5118
},
{
"epoch": 1.7852572484613909,
"grad_norm": 0.16943421959877014,
"learning_rate": 1.2806143628061437e-05,
"loss": 1.8783817291259766,
"step": 5120
},
{
"epoch": 1.7859546350053175,
"grad_norm": 0.2601114809513092,
"learning_rate": 1.2764632627646327e-05,
"loss": 1.792428731918335,
"step": 5122
},
{
"epoch": 1.7866520215492443,
"grad_norm": 0.21732443571090698,
"learning_rate": 1.2723121627231215e-05,
"loss": 1.6711856126785278,
"step": 5124
},
{
"epoch": 1.7873494080931709,
"grad_norm": 0.2170199751853943,
"learning_rate": 1.2681610626816107e-05,
"loss": 1.8425421714782715,
"step": 5126
},
{
"epoch": 1.7880467946370975,
"grad_norm": 0.17884907126426697,
"learning_rate": 1.2640099626400997e-05,
"loss": 1.9479386806488037,
"step": 5128
},
{
"epoch": 1.788744181181024,
"grad_norm": 0.17128115892410278,
"learning_rate": 1.2598588625985888e-05,
"loss": 1.821108102798462,
"step": 5130
},
{
"epoch": 1.7894415677249507,
"grad_norm": 0.24064293503761292,
"learning_rate": 1.2557077625570777e-05,
"loss": 1.6873881816864014,
"step": 5132
},
{
"epoch": 1.7901389542688775,
"grad_norm": 0.3249698579311371,
"learning_rate": 1.2515566625155666e-05,
"loss": 1.8007752895355225,
"step": 5134
},
{
"epoch": 1.7908363408128039,
"grad_norm": 0.24538260698318481,
"learning_rate": 1.2474055624740556e-05,
"loss": 1.9702765941619873,
"step": 5136
},
{
"epoch": 1.7915337273567307,
"grad_norm": 0.21633081138134003,
"learning_rate": 1.2432544624325446e-05,
"loss": 1.8523898124694824,
"step": 5138
},
{
"epoch": 1.7922311139006573,
"grad_norm": 0.2477222979068756,
"learning_rate": 1.2391033623910336e-05,
"loss": 1.9169845581054688,
"step": 5140
},
{
"epoch": 1.7929285004445839,
"grad_norm": 0.22729334235191345,
"learning_rate": 1.2349522623495228e-05,
"loss": 1.6350218057632446,
"step": 5142
},
{
"epoch": 1.7936258869885107,
"grad_norm": 0.1738540232181549,
"learning_rate": 1.2308011623080116e-05,
"loss": 1.9938302040100098,
"step": 5144
},
{
"epoch": 1.794323273532437,
"grad_norm": 0.20553775131702423,
"learning_rate": 1.2266500622665007e-05,
"loss": 1.8777868747711182,
"step": 5146
},
{
"epoch": 1.7950206600763638,
"grad_norm": 0.17821897566318512,
"learning_rate": 1.2224989622249897e-05,
"loss": 1.9826552867889404,
"step": 5148
},
{
"epoch": 1.7957180466202904,
"grad_norm": 0.20037521421909332,
"learning_rate": 1.2183478621834787e-05,
"loss": 1.978858470916748,
"step": 5150
},
{
"epoch": 1.796415433164217,
"grad_norm": 0.2013157457113266,
"learning_rate": 1.2141967621419677e-05,
"loss": 1.7920923233032227,
"step": 5152
},
{
"epoch": 1.7971128197081438,
"grad_norm": 0.24478378891944885,
"learning_rate": 1.2100456621004567e-05,
"loss": 1.81707763671875,
"step": 5154
},
{
"epoch": 1.7978102062520702,
"grad_norm": 0.23601314425468445,
"learning_rate": 1.2058945620589457e-05,
"loss": 1.911826729774475,
"step": 5156
},
{
"epoch": 1.798507592795997,
"grad_norm": 0.21808622777462006,
"learning_rate": 1.2017434620174347e-05,
"loss": 1.8675727844238281,
"step": 5158
},
{
"epoch": 1.7992049793399236,
"grad_norm": 0.17870546877384186,
"learning_rate": 1.1975923619759236e-05,
"loss": 1.9483091831207275,
"step": 5160
},
{
"epoch": 1.7999023658838502,
"grad_norm": 0.22030431032180786,
"learning_rate": 1.1934412619344126e-05,
"loss": 2.0492300987243652,
"step": 5162
},
{
"epoch": 1.800599752427777,
"grad_norm": 0.22699806094169617,
"learning_rate": 1.1892901618929018e-05,
"loss": 1.799140214920044,
"step": 5164
},
{
"epoch": 1.8012971389717034,
"grad_norm": 0.2665795683860779,
"learning_rate": 1.1851390618513906e-05,
"loss": 1.7880096435546875,
"step": 5166
}
],
"logging_steps": 2,
"max_steps": 5736,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 574,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 9.246842507776492e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}