sparky-decoder-3 / checkpoint-500 /trainer_state.json
Dohahemdann's picture
Checkpoint checkpoint-500
1c5d2b9 verified
Raw
History Blame Contribute Delete
86.7 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.4207868714496108,
"eval_steps": 500,
"global_step": 500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0008415737428992215,
"grad_norm": 1220866.75,
"learning_rate": 0.0,
"loss": 6.7045,
"step": 1
},
{
"epoch": 0.001683147485798443,
"grad_norm": 1159900.75,
"learning_rate": 2.0000000000000002e-07,
"loss": 5.5719,
"step": 2
},
{
"epoch": 0.0025247212286976647,
"grad_norm": 141534.25,
"learning_rate": 4.0000000000000003e-07,
"loss": 2.934,
"step": 3
},
{
"epoch": 0.003366294971596886,
"grad_norm": 46414.40625,
"learning_rate": 6.000000000000001e-07,
"loss": 2.3661,
"step": 4
},
{
"epoch": 0.004207868714496108,
"grad_norm": 24856.083984375,
"learning_rate": 8.000000000000001e-07,
"loss": 2.4011,
"step": 5
},
{
"epoch": 0.005049442457395329,
"grad_norm": 12771.7001953125,
"learning_rate": 1.0000000000000002e-06,
"loss": 2.0539,
"step": 6
},
{
"epoch": 0.005891016200294551,
"grad_norm": 11572.16015625,
"learning_rate": 1.2000000000000002e-06,
"loss": 2.0979,
"step": 7
},
{
"epoch": 0.006732589943193772,
"grad_norm": 6822.94873046875,
"learning_rate": 1.4000000000000001e-06,
"loss": 2.0854,
"step": 8
},
{
"epoch": 0.007574163686092994,
"grad_norm": 7443.47412109375,
"learning_rate": 1.6000000000000001e-06,
"loss": 2.4363,
"step": 9
},
{
"epoch": 0.008415737428992216,
"grad_norm": 4867.23095703125,
"learning_rate": 1.8000000000000001e-06,
"loss": 2.1086,
"step": 10
},
{
"epoch": 0.009257311171891438,
"grad_norm": 4013.554443359375,
"learning_rate": 2.0000000000000003e-06,
"loss": 1.7879,
"step": 11
},
{
"epoch": 0.010098884914790659,
"grad_norm": 3542.235107421875,
"learning_rate": 2.2e-06,
"loss": 1.6868,
"step": 12
},
{
"epoch": 0.01094045865768988,
"grad_norm": 3830.751220703125,
"learning_rate": 2.4000000000000003e-06,
"loss": 1.7032,
"step": 13
},
{
"epoch": 0.011782032400589101,
"grad_norm": 2990.0068359375,
"learning_rate": 2.6e-06,
"loss": 1.6582,
"step": 14
},
{
"epoch": 0.012623606143488323,
"grad_norm": 3060.302001953125,
"learning_rate": 2.8000000000000003e-06,
"loss": 1.8881,
"step": 15
},
{
"epoch": 0.013465179886387544,
"grad_norm": 3624.901611328125,
"learning_rate": 3e-06,
"loss": 1.5175,
"step": 16
},
{
"epoch": 0.014306753629286767,
"grad_norm": 2495.706298828125,
"learning_rate": 3.2000000000000003e-06,
"loss": 1.8591,
"step": 17
},
{
"epoch": 0.015148327372185988,
"grad_norm": 4080.043701171875,
"learning_rate": 3.4000000000000005e-06,
"loss": 1.9858,
"step": 18
},
{
"epoch": 0.01598990111508521,
"grad_norm": 2687.82470703125,
"learning_rate": 3.6000000000000003e-06,
"loss": 1.8581,
"step": 19
},
{
"epoch": 0.016831474857984433,
"grad_norm": 3068.358154296875,
"learning_rate": 3.8000000000000005e-06,
"loss": 1.8357,
"step": 20
},
{
"epoch": 0.017673048600883652,
"grad_norm": 1605.349609375,
"learning_rate": 4.000000000000001e-06,
"loss": 1.5524,
"step": 21
},
{
"epoch": 0.018514622343782875,
"grad_norm": 2444.8876953125,
"learning_rate": 4.2000000000000004e-06,
"loss": 1.4551,
"step": 22
},
{
"epoch": 0.019356196086682095,
"grad_norm": 1613.062744140625,
"learning_rate": 4.4e-06,
"loss": 1.6651,
"step": 23
},
{
"epoch": 0.020197769829581318,
"grad_norm": 1513.2115478515625,
"learning_rate": 4.600000000000001e-06,
"loss": 1.7729,
"step": 24
},
{
"epoch": 0.021039343572480537,
"grad_norm": 2948.156982421875,
"learning_rate": 4.800000000000001e-06,
"loss": 2.2835,
"step": 25
},
{
"epoch": 0.02188091731537976,
"grad_norm": 1072.2357177734375,
"learning_rate": 5e-06,
"loss": 1.6071,
"step": 26
},
{
"epoch": 0.022722491058278983,
"grad_norm": 1390.9981689453125,
"learning_rate": 5.2e-06,
"loss": 1.6213,
"step": 27
},
{
"epoch": 0.023564064801178203,
"grad_norm": 1082.24169921875,
"learning_rate": 5.400000000000001e-06,
"loss": 1.6285,
"step": 28
},
{
"epoch": 0.024405638544077426,
"grad_norm": 1373.3822021484375,
"learning_rate": 5.600000000000001e-06,
"loss": 1.7561,
"step": 29
},
{
"epoch": 0.025247212286976645,
"grad_norm": 990.3181762695312,
"learning_rate": 5.8e-06,
"loss": 1.6592,
"step": 30
},
{
"epoch": 0.02608878602987587,
"grad_norm": 1283.081298828125,
"learning_rate": 6e-06,
"loss": 1.5235,
"step": 31
},
{
"epoch": 0.026930359772775088,
"grad_norm": 1171.0947265625,
"learning_rate": 6.200000000000001e-06,
"loss": 1.5765,
"step": 32
},
{
"epoch": 0.02777193351567431,
"grad_norm": 1270.686767578125,
"learning_rate": 6.4000000000000006e-06,
"loss": 1.7905,
"step": 33
},
{
"epoch": 0.028613507258573534,
"grad_norm": 962.5809326171875,
"learning_rate": 6.600000000000001e-06,
"loss": 1.4895,
"step": 34
},
{
"epoch": 0.029455081001472753,
"grad_norm": 747.169677734375,
"learning_rate": 6.800000000000001e-06,
"loss": 1.1956,
"step": 35
},
{
"epoch": 0.030296654744371977,
"grad_norm": 937.03759765625,
"learning_rate": 7e-06,
"loss": 1.345,
"step": 36
},
{
"epoch": 0.031138228487271196,
"grad_norm": 814.0247802734375,
"learning_rate": 7.2000000000000005e-06,
"loss": 1.4555,
"step": 37
},
{
"epoch": 0.03197980223017042,
"grad_norm": 1292.210693359375,
"learning_rate": 7.4e-06,
"loss": 1.5723,
"step": 38
},
{
"epoch": 0.03282137597306964,
"grad_norm": 1027.9407958984375,
"learning_rate": 7.600000000000001e-06,
"loss": 1.4792,
"step": 39
},
{
"epoch": 0.033662949715968865,
"grad_norm": 848.7986450195312,
"learning_rate": 7.800000000000002e-06,
"loss": 1.2985,
"step": 40
},
{
"epoch": 0.034504523458868085,
"grad_norm": 890.843017578125,
"learning_rate": 8.000000000000001e-06,
"loss": 1.5241,
"step": 41
},
{
"epoch": 0.035346097201767304,
"grad_norm": 712.9835205078125,
"learning_rate": 8.2e-06,
"loss": 1.6666,
"step": 42
},
{
"epoch": 0.036187670944666524,
"grad_norm": 744.9312133789062,
"learning_rate": 8.400000000000001e-06,
"loss": 1.8666,
"step": 43
},
{
"epoch": 0.03702924468756575,
"grad_norm": 698.9174194335938,
"learning_rate": 8.6e-06,
"loss": 1.6878,
"step": 44
},
{
"epoch": 0.03787081843046497,
"grad_norm": 541.9822998046875,
"learning_rate": 8.8e-06,
"loss": 1.1321,
"step": 45
},
{
"epoch": 0.03871239217336419,
"grad_norm": 566.4813842773438,
"learning_rate": 9e-06,
"loss": 1.7059,
"step": 46
},
{
"epoch": 0.039553965916263416,
"grad_norm": 804.7236938476562,
"learning_rate": 9.200000000000002e-06,
"loss": 1.4996,
"step": 47
},
{
"epoch": 0.040395539659162635,
"grad_norm": 779.8905029296875,
"learning_rate": 9.4e-06,
"loss": 1.4958,
"step": 48
},
{
"epoch": 0.041237113402061855,
"grad_norm": 707.608642578125,
"learning_rate": 9.600000000000001e-06,
"loss": 1.6273,
"step": 49
},
{
"epoch": 0.042078687144961074,
"grad_norm": 718.4205932617188,
"learning_rate": 9.800000000000001e-06,
"loss": 1.6194,
"step": 50
},
{
"epoch": 0.0429202608878603,
"grad_norm": 650.2810668945312,
"learning_rate": 1e-05,
"loss": 1.6353,
"step": 51
},
{
"epoch": 0.04376183463075952,
"grad_norm": 576.8077392578125,
"learning_rate": 9.999995447250702e-06,
"loss": 1.5377,
"step": 52
},
{
"epoch": 0.04460340837365874,
"grad_norm": 1010.9555053710938,
"learning_rate": 9.999981789011094e-06,
"loss": 1.5717,
"step": 53
},
{
"epoch": 0.045444982116557966,
"grad_norm": 485.38592529296875,
"learning_rate": 9.999959025306055e-06,
"loss": 1.4595,
"step": 54
},
{
"epoch": 0.046286555859457186,
"grad_norm": 446.7446594238281,
"learning_rate": 9.999927156177032e-06,
"loss": 1.65,
"step": 55
},
{
"epoch": 0.047128129602356406,
"grad_norm": 501.9324951171875,
"learning_rate": 9.999886181682071e-06,
"loss": 1.4112,
"step": 56
},
{
"epoch": 0.047969703345255625,
"grad_norm": 659.2301635742188,
"learning_rate": 9.999836101895782e-06,
"loss": 1.4711,
"step": 57
},
{
"epoch": 0.04881127708815485,
"grad_norm": 368.6539001464844,
"learning_rate": 9.999776916909372e-06,
"loss": 1.4114,
"step": 58
},
{
"epoch": 0.04965285083105407,
"grad_norm": 374.8740539550781,
"learning_rate": 9.999708626830617e-06,
"loss": 1.1645,
"step": 59
},
{
"epoch": 0.05049442457395329,
"grad_norm": 632.1150512695312,
"learning_rate": 9.999631231783886e-06,
"loss": 1.753,
"step": 60
},
{
"epoch": 0.05133599831685252,
"grad_norm": 387.0769348144531,
"learning_rate": 9.99954473191012e-06,
"loss": 1.4465,
"step": 61
},
{
"epoch": 0.05217757205975174,
"grad_norm": 344.7577819824219,
"learning_rate": 9.999449127366842e-06,
"loss": 1.3878,
"step": 62
},
{
"epoch": 0.053019145802650956,
"grad_norm": 347.4836120605469,
"learning_rate": 9.999344418328161e-06,
"loss": 1.2818,
"step": 63
},
{
"epoch": 0.053860719545550176,
"grad_norm": 704.4097900390625,
"learning_rate": 9.999230604984763e-06,
"loss": 1.3777,
"step": 64
},
{
"epoch": 0.0547022932884494,
"grad_norm": 465.4077453613281,
"learning_rate": 9.999107687543907e-06,
"loss": 1.6164,
"step": 65
},
{
"epoch": 0.05554386703134862,
"grad_norm": 460.2505187988281,
"learning_rate": 9.998975666229445e-06,
"loss": 1.4093,
"step": 66
},
{
"epoch": 0.05638544077424784,
"grad_norm": 414.6778869628906,
"learning_rate": 9.998834541281798e-06,
"loss": 1.2764,
"step": 67
},
{
"epoch": 0.05722701451714707,
"grad_norm": 410.7452697753906,
"learning_rate": 9.998684312957972e-06,
"loss": 1.3213,
"step": 68
},
{
"epoch": 0.05806858826004629,
"grad_norm": 454.0287780761719,
"learning_rate": 9.99852498153154e-06,
"loss": 1.6469,
"step": 69
},
{
"epoch": 0.05891016200294551,
"grad_norm": 382.1865539550781,
"learning_rate": 9.998356547292668e-06,
"loss": 1.5022,
"step": 70
},
{
"epoch": 0.059751735745844727,
"grad_norm": 527.5244750976562,
"learning_rate": 9.998179010548087e-06,
"loss": 1.3488,
"step": 71
},
{
"epoch": 0.06059330948874395,
"grad_norm": 667.3890380859375,
"learning_rate": 9.997992371621114e-06,
"loss": 1.4324,
"step": 72
},
{
"epoch": 0.06143488323164317,
"grad_norm": 271.77227783203125,
"learning_rate": 9.997796630851632e-06,
"loss": 1.2182,
"step": 73
},
{
"epoch": 0.06227645697454239,
"grad_norm": 285.75909423828125,
"learning_rate": 9.997591788596106e-06,
"loss": 1.6096,
"step": 74
},
{
"epoch": 0.06311803071744161,
"grad_norm": 446.6322021484375,
"learning_rate": 9.997377845227577e-06,
"loss": 1.3402,
"step": 75
},
{
"epoch": 0.06395960446034084,
"grad_norm": 265.3509216308594,
"learning_rate": 9.997154801135651e-06,
"loss": 1.4986,
"step": 76
},
{
"epoch": 0.06480117820324006,
"grad_norm": 307.3344421386719,
"learning_rate": 9.99692265672652e-06,
"loss": 1.39,
"step": 77
},
{
"epoch": 0.06564275194613928,
"grad_norm": 272.6602478027344,
"learning_rate": 9.996681412422939e-06,
"loss": 1.2164,
"step": 78
},
{
"epoch": 0.0664843256890385,
"grad_norm": 279.1375732421875,
"learning_rate": 9.996431068664237e-06,
"loss": 1.0068,
"step": 79
},
{
"epoch": 0.06732589943193773,
"grad_norm": 446.9021911621094,
"learning_rate": 9.996171625906318e-06,
"loss": 1.5697,
"step": 80
},
{
"epoch": 0.06816747317483694,
"grad_norm": 335.849853515625,
"learning_rate": 9.995903084621651e-06,
"loss": 1.5072,
"step": 81
},
{
"epoch": 0.06900904691773617,
"grad_norm": 286.28125,
"learning_rate": 9.995625445299277e-06,
"loss": 1.6705,
"step": 82
},
{
"epoch": 0.06985062066063538,
"grad_norm": 258.25634765625,
"learning_rate": 9.995338708444804e-06,
"loss": 1.2367,
"step": 83
},
{
"epoch": 0.07069219440353461,
"grad_norm": 411.6665954589844,
"learning_rate": 9.99504287458041e-06,
"loss": 1.1231,
"step": 84
},
{
"epoch": 0.07153376814643383,
"grad_norm": 208.81202697753906,
"learning_rate": 9.994737944244837e-06,
"loss": 1.0963,
"step": 85
},
{
"epoch": 0.07237534188933305,
"grad_norm": 285.0477294921875,
"learning_rate": 9.994423917993395e-06,
"loss": 1.2332,
"step": 86
},
{
"epoch": 0.07321691563223227,
"grad_norm": 228.11558532714844,
"learning_rate": 9.994100796397954e-06,
"loss": 1.3684,
"step": 87
},
{
"epoch": 0.0740584893751315,
"grad_norm": 311.3058776855469,
"learning_rate": 9.993768580046954e-06,
"loss": 1.4219,
"step": 88
},
{
"epoch": 0.07490006311803071,
"grad_norm": 191.33192443847656,
"learning_rate": 9.993427269545391e-06,
"loss": 0.9917,
"step": 89
},
{
"epoch": 0.07574163686092994,
"grad_norm": 252.05970764160156,
"learning_rate": 9.993076865514829e-06,
"loss": 1.2707,
"step": 90
},
{
"epoch": 0.07658321060382917,
"grad_norm": 209.5124053955078,
"learning_rate": 9.992717368593385e-06,
"loss": 1.4056,
"step": 91
},
{
"epoch": 0.07742478434672838,
"grad_norm": 306.5719299316406,
"learning_rate": 9.992348779435743e-06,
"loss": 1.3482,
"step": 92
},
{
"epoch": 0.0782663580896276,
"grad_norm": 232.7973175048828,
"learning_rate": 9.991971098713136e-06,
"loss": 1.2889,
"step": 93
},
{
"epoch": 0.07910793183252683,
"grad_norm": 345.8323974609375,
"learning_rate": 9.99158432711336e-06,
"loss": 1.2533,
"step": 94
},
{
"epoch": 0.07994950557542604,
"grad_norm": 252.8789520263672,
"learning_rate": 9.991188465340766e-06,
"loss": 1.0905,
"step": 95
},
{
"epoch": 0.08079107931832527,
"grad_norm": 192.878662109375,
"learning_rate": 9.990783514116257e-06,
"loss": 1.0254,
"step": 96
},
{
"epoch": 0.08163265306122448,
"grad_norm": 174.9850311279297,
"learning_rate": 9.990369474177287e-06,
"loss": 1.1414,
"step": 97
},
{
"epoch": 0.08247422680412371,
"grad_norm": 237.68482971191406,
"learning_rate": 9.98994634627787e-06,
"loss": 1.392,
"step": 98
},
{
"epoch": 0.08331580054702294,
"grad_norm": 254.39601135253906,
"learning_rate": 9.98951413118856e-06,
"loss": 1.7766,
"step": 99
},
{
"epoch": 0.08415737428992215,
"grad_norm": 280.639892578125,
"learning_rate": 9.989072829696462e-06,
"loss": 1.1928,
"step": 100
},
{
"epoch": 0.08499894803282138,
"grad_norm": 228.63319396972656,
"learning_rate": 9.988622442605233e-06,
"loss": 1.1746,
"step": 101
},
{
"epoch": 0.0858405217757206,
"grad_norm": 214.7047882080078,
"learning_rate": 9.988162970735073e-06,
"loss": 1.2603,
"step": 102
},
{
"epoch": 0.08668209551861981,
"grad_norm": 202.94602966308594,
"learning_rate": 9.987694414922724e-06,
"loss": 1.0886,
"step": 103
},
{
"epoch": 0.08752366926151904,
"grad_norm": 199.387939453125,
"learning_rate": 9.987216776021476e-06,
"loss": 1.0365,
"step": 104
},
{
"epoch": 0.08836524300441827,
"grad_norm": 213.04771423339844,
"learning_rate": 9.986730054901153e-06,
"loss": 1.3704,
"step": 105
},
{
"epoch": 0.08920681674731748,
"grad_norm": 205.0407257080078,
"learning_rate": 9.986234252448127e-06,
"loss": 1.234,
"step": 106
},
{
"epoch": 0.0900483904902167,
"grad_norm": 210.17430114746094,
"learning_rate": 9.985729369565299e-06,
"loss": 1.2648,
"step": 107
},
{
"epoch": 0.09088996423311593,
"grad_norm": 237.6966552734375,
"learning_rate": 9.985215407172116e-06,
"loss": 1.3181,
"step": 108
},
{
"epoch": 0.09173153797601515,
"grad_norm": 221.96646118164062,
"learning_rate": 9.98469236620455e-06,
"loss": 1.3234,
"step": 109
},
{
"epoch": 0.09257311171891437,
"grad_norm": 170.47254943847656,
"learning_rate": 9.984160247615114e-06,
"loss": 1.0807,
"step": 110
},
{
"epoch": 0.09341468546181358,
"grad_norm": 251.4197235107422,
"learning_rate": 9.983619052372847e-06,
"loss": 1.0354,
"step": 111
},
{
"epoch": 0.09425625920471281,
"grad_norm": 152.14559936523438,
"learning_rate": 9.983068781463324e-06,
"loss": 1.1532,
"step": 112
},
{
"epoch": 0.09509783294761204,
"grad_norm": 226.23793029785156,
"learning_rate": 9.982509435888636e-06,
"loss": 1.1527,
"step": 113
},
{
"epoch": 0.09593940669051125,
"grad_norm": 169.29112243652344,
"learning_rate": 9.981941016667413e-06,
"loss": 1.2584,
"step": 114
},
{
"epoch": 0.09678098043341048,
"grad_norm": 165.5533905029297,
"learning_rate": 9.9813635248348e-06,
"loss": 1.249,
"step": 115
},
{
"epoch": 0.0976225541763097,
"grad_norm": 156.0625,
"learning_rate": 9.980776961442469e-06,
"loss": 1.0096,
"step": 116
},
{
"epoch": 0.09846412791920892,
"grad_norm": 147.66213989257812,
"learning_rate": 9.98018132755861e-06,
"loss": 1.2037,
"step": 117
},
{
"epoch": 0.09930570166210814,
"grad_norm": 196.01304626464844,
"learning_rate": 9.97957662426793e-06,
"loss": 1.3726,
"step": 118
},
{
"epoch": 0.10014727540500737,
"grad_norm": 140.2161865234375,
"learning_rate": 9.978962852671656e-06,
"loss": 1.0369,
"step": 119
},
{
"epoch": 0.10098884914790658,
"grad_norm": 197.426513671875,
"learning_rate": 9.978340013887527e-06,
"loss": 1.1255,
"step": 120
},
{
"epoch": 0.10183042289080581,
"grad_norm": 186.78482055664062,
"learning_rate": 9.977708109049794e-06,
"loss": 1.1633,
"step": 121
},
{
"epoch": 0.10267199663370503,
"grad_norm": 169.86045837402344,
"learning_rate": 9.977067139309218e-06,
"loss": 1.4329,
"step": 122
},
{
"epoch": 0.10351357037660425,
"grad_norm": 190.2769317626953,
"learning_rate": 9.97641710583307e-06,
"loss": 1.1239,
"step": 123
},
{
"epoch": 0.10435514411950347,
"grad_norm": 257.3651428222656,
"learning_rate": 9.975758009805127e-06,
"loss": 0.9102,
"step": 124
},
{
"epoch": 0.1051967178624027,
"grad_norm": 149.8079071044922,
"learning_rate": 9.975089852425666e-06,
"loss": 1.034,
"step": 125
},
{
"epoch": 0.10603829160530191,
"grad_norm": 140.86318969726562,
"learning_rate": 9.974412634911467e-06,
"loss": 1.0696,
"step": 126
},
{
"epoch": 0.10687986534820114,
"grad_norm": 142.2266387939453,
"learning_rate": 9.973726358495816e-06,
"loss": 1.1112,
"step": 127
},
{
"epoch": 0.10772143909110035,
"grad_norm": 161.84036254882812,
"learning_rate": 9.973031024428487e-06,
"loss": 1.3477,
"step": 128
},
{
"epoch": 0.10856301283399958,
"grad_norm": 122.45903015136719,
"learning_rate": 9.972326633975754e-06,
"loss": 0.9899,
"step": 129
},
{
"epoch": 0.1094045865768988,
"grad_norm": 154.53138732910156,
"learning_rate": 9.971613188420379e-06,
"loss": 1.0933,
"step": 130
},
{
"epoch": 0.11024616031979802,
"grad_norm": 146.3953399658203,
"learning_rate": 9.970890689061622e-06,
"loss": 0.9027,
"step": 131
},
{
"epoch": 0.11108773406269724,
"grad_norm": 137.6485595703125,
"learning_rate": 9.970159137215225e-06,
"loss": 1.034,
"step": 132
},
{
"epoch": 0.11192930780559647,
"grad_norm": 218.57345581054688,
"learning_rate": 9.969418534213415e-06,
"loss": 1.171,
"step": 133
},
{
"epoch": 0.11277088154849568,
"grad_norm": 164.52330017089844,
"learning_rate": 9.968668881404906e-06,
"loss": 1.2049,
"step": 134
},
{
"epoch": 0.11361245529139491,
"grad_norm": 124.15032958984375,
"learning_rate": 9.96791018015489e-06,
"loss": 1.2256,
"step": 135
},
{
"epoch": 0.11445402903429414,
"grad_norm": 174.1881561279297,
"learning_rate": 9.967142431845037e-06,
"loss": 1.2492,
"step": 136
},
{
"epoch": 0.11529560277719335,
"grad_norm": 116.0736083984375,
"learning_rate": 9.966365637873493e-06,
"loss": 0.9083,
"step": 137
},
{
"epoch": 0.11613717652009257,
"grad_norm": 161.2622528076172,
"learning_rate": 9.965579799654878e-06,
"loss": 1.1455,
"step": 138
},
{
"epoch": 0.1169787502629918,
"grad_norm": 154.3689422607422,
"learning_rate": 9.964784918620284e-06,
"loss": 1.3865,
"step": 139
},
{
"epoch": 0.11782032400589101,
"grad_norm": 120.22959899902344,
"learning_rate": 9.963980996217264e-06,
"loss": 0.8486,
"step": 140
},
{
"epoch": 0.11866189774879024,
"grad_norm": 107.49683380126953,
"learning_rate": 9.963168033909843e-06,
"loss": 1.1682,
"step": 141
},
{
"epoch": 0.11950347149168945,
"grad_norm": 110.22765350341797,
"learning_rate": 9.962346033178508e-06,
"loss": 0.9903,
"step": 142
},
{
"epoch": 0.12034504523458868,
"grad_norm": 145.84603881835938,
"learning_rate": 9.961514995520202e-06,
"loss": 1.1753,
"step": 143
},
{
"epoch": 0.1211866189774879,
"grad_norm": 124.43152618408203,
"learning_rate": 9.960674922448327e-06,
"loss": 1.1763,
"step": 144
},
{
"epoch": 0.12202819272038712,
"grad_norm": 123.75733184814453,
"learning_rate": 9.959825815492745e-06,
"loss": 1.127,
"step": 145
},
{
"epoch": 0.12286976646328635,
"grad_norm": 142.31790161132812,
"learning_rate": 9.958967676199758e-06,
"loss": 1.0565,
"step": 146
},
{
"epoch": 0.12371134020618557,
"grad_norm": 121.63768768310547,
"learning_rate": 9.958100506132127e-06,
"loss": 1.0754,
"step": 147
},
{
"epoch": 0.12455291394908478,
"grad_norm": 119.10518646240234,
"learning_rate": 9.957224306869053e-06,
"loss": 1.0634,
"step": 148
},
{
"epoch": 0.125394487691984,
"grad_norm": 123.80194854736328,
"learning_rate": 9.956339080006184e-06,
"loss": 1.2764,
"step": 149
},
{
"epoch": 0.12623606143488322,
"grad_norm": 127.6620864868164,
"learning_rate": 9.955444827155605e-06,
"loss": 1.0116,
"step": 150
},
{
"epoch": 0.12707763517778245,
"grad_norm": 125.85498809814453,
"learning_rate": 9.954541549945841e-06,
"loss": 1.0106,
"step": 151
},
{
"epoch": 0.12791920892068168,
"grad_norm": 187.7716827392578,
"learning_rate": 9.953629250021847e-06,
"loss": 0.967,
"step": 152
},
{
"epoch": 0.1287607826635809,
"grad_norm": 113.58953094482422,
"learning_rate": 9.952707929045018e-06,
"loss": 0.9723,
"step": 153
},
{
"epoch": 0.12960235640648013,
"grad_norm": 124.91770935058594,
"learning_rate": 9.951777588693166e-06,
"loss": 0.833,
"step": 154
},
{
"epoch": 0.13044393014937933,
"grad_norm": 99.47856903076172,
"learning_rate": 9.950838230660535e-06,
"loss": 0.9527,
"step": 155
},
{
"epoch": 0.13128550389227855,
"grad_norm": 163.18048095703125,
"learning_rate": 9.949889856657789e-06,
"loss": 1.0965,
"step": 156
},
{
"epoch": 0.13212707763517778,
"grad_norm": 120.51565551757812,
"learning_rate": 9.948932468412013e-06,
"loss": 0.9404,
"step": 157
},
{
"epoch": 0.132968651378077,
"grad_norm": 140.493896484375,
"learning_rate": 9.947966067666707e-06,
"loss": 0.9947,
"step": 158
},
{
"epoch": 0.13381022512097623,
"grad_norm": 143.33267211914062,
"learning_rate": 9.946990656181782e-06,
"loss": 1.0438,
"step": 159
},
{
"epoch": 0.13465179886387546,
"grad_norm": 152.41970825195312,
"learning_rate": 9.94600623573356e-06,
"loss": 1.2963,
"step": 160
},
{
"epoch": 0.13549337260677466,
"grad_norm": 142.91018676757812,
"learning_rate": 9.945012808114765e-06,
"loss": 0.8879,
"step": 161
},
{
"epoch": 0.13633494634967389,
"grad_norm": 136.81448364257812,
"learning_rate": 9.944010375134533e-06,
"loss": 1.1252,
"step": 162
},
{
"epoch": 0.1371765200925731,
"grad_norm": 172.33338928222656,
"learning_rate": 9.942998938618394e-06,
"loss": 1.0413,
"step": 163
},
{
"epoch": 0.13801809383547234,
"grad_norm": 100.11540222167969,
"learning_rate": 9.941978500408273e-06,
"loss": 1.017,
"step": 164
},
{
"epoch": 0.13885966757837157,
"grad_norm": 188.677490234375,
"learning_rate": 9.94094906236249e-06,
"loss": 1.129,
"step": 165
},
{
"epoch": 0.13970124132127076,
"grad_norm": 135.1410675048828,
"learning_rate": 9.939910626355754e-06,
"loss": 1.2248,
"step": 166
},
{
"epoch": 0.14054281506417,
"grad_norm": 160.2061309814453,
"learning_rate": 9.938863194279163e-06,
"loss": 1.2495,
"step": 167
},
{
"epoch": 0.14138438880706922,
"grad_norm": 140.02020263671875,
"learning_rate": 9.93780676804019e-06,
"loss": 1.208,
"step": 168
},
{
"epoch": 0.14222596254996844,
"grad_norm": 118.90058135986328,
"learning_rate": 9.936741349562698e-06,
"loss": 0.8609,
"step": 169
},
{
"epoch": 0.14306753629286767,
"grad_norm": 117.95634460449219,
"learning_rate": 9.935666940786917e-06,
"loss": 1.151,
"step": 170
},
{
"epoch": 0.1439091100357669,
"grad_norm": 350.86871337890625,
"learning_rate": 9.934583543669454e-06,
"loss": 1.0442,
"step": 171
},
{
"epoch": 0.1447506837786661,
"grad_norm": 123.0718002319336,
"learning_rate": 9.933491160183283e-06,
"loss": 1.1908,
"step": 172
},
{
"epoch": 0.14559225752156532,
"grad_norm": 128.88368225097656,
"learning_rate": 9.932389792317743e-06,
"loss": 0.9598,
"step": 173
},
{
"epoch": 0.14643383126446455,
"grad_norm": 106.69990539550781,
"learning_rate": 9.931279442078533e-06,
"loss": 1.0026,
"step": 174
},
{
"epoch": 0.14727540500736377,
"grad_norm": 100.58325958251953,
"learning_rate": 9.930160111487717e-06,
"loss": 1.0398,
"step": 175
},
{
"epoch": 0.148116978750263,
"grad_norm": 105.3095474243164,
"learning_rate": 9.929031802583702e-06,
"loss": 1.1148,
"step": 176
},
{
"epoch": 0.1489585524931622,
"grad_norm": 134.05030822753906,
"learning_rate": 9.927894517421252e-06,
"loss": 0.9044,
"step": 177
},
{
"epoch": 0.14980012623606143,
"grad_norm": 142.59539794921875,
"learning_rate": 9.92674825807148e-06,
"loss": 1.0141,
"step": 178
},
{
"epoch": 0.15064169997896065,
"grad_norm": 76.3344497680664,
"learning_rate": 9.925593026621833e-06,
"loss": 0.8618,
"step": 179
},
{
"epoch": 0.15148327372185988,
"grad_norm": 108.6050796508789,
"learning_rate": 9.924428825176107e-06,
"loss": 1.1011,
"step": 180
},
{
"epoch": 0.1523248474647591,
"grad_norm": 81.81193542480469,
"learning_rate": 9.923255655854428e-06,
"loss": 0.7179,
"step": 181
},
{
"epoch": 0.15316642120765833,
"grad_norm": 101.39788055419922,
"learning_rate": 9.922073520793255e-06,
"loss": 1.2286,
"step": 182
},
{
"epoch": 0.15400799495055753,
"grad_norm": 106.66728973388672,
"learning_rate": 9.920882422145372e-06,
"loss": 0.988,
"step": 183
},
{
"epoch": 0.15484956869345676,
"grad_norm": 107.3541030883789,
"learning_rate": 9.91968236207989e-06,
"loss": 0.9532,
"step": 184
},
{
"epoch": 0.15569114243635598,
"grad_norm": 88.04878997802734,
"learning_rate": 9.918473342782236e-06,
"loss": 0.7913,
"step": 185
},
{
"epoch": 0.1565327161792552,
"grad_norm": 98.0555191040039,
"learning_rate": 9.917255366454157e-06,
"loss": 1.073,
"step": 186
},
{
"epoch": 0.15737428992215444,
"grad_norm": 83.71012878417969,
"learning_rate": 9.91602843531371e-06,
"loss": 0.848,
"step": 187
},
{
"epoch": 0.15821586366505366,
"grad_norm": 98.08084869384766,
"learning_rate": 9.914792551595255e-06,
"loss": 0.883,
"step": 188
},
{
"epoch": 0.15905743740795286,
"grad_norm": 80.48824310302734,
"learning_rate": 9.913547717549463e-06,
"loss": 0.9977,
"step": 189
},
{
"epoch": 0.1598990111508521,
"grad_norm": 107.66862487792969,
"learning_rate": 9.9122939354433e-06,
"loss": 1.0573,
"step": 190
},
{
"epoch": 0.16074058489375131,
"grad_norm": 89.83882904052734,
"learning_rate": 9.911031207560027e-06,
"loss": 1.0187,
"step": 191
},
{
"epoch": 0.16158215863665054,
"grad_norm": 100.37699890136719,
"learning_rate": 9.909759536199199e-06,
"loss": 1.1422,
"step": 192
},
{
"epoch": 0.16242373237954977,
"grad_norm": 86.7760009765625,
"learning_rate": 9.908478923676657e-06,
"loss": 1.0619,
"step": 193
},
{
"epoch": 0.16326530612244897,
"grad_norm": 82.6327896118164,
"learning_rate": 9.907189372324522e-06,
"loss": 1.0063,
"step": 194
},
{
"epoch": 0.1641068798653482,
"grad_norm": 78.38849639892578,
"learning_rate": 9.905890884491196e-06,
"loss": 1.0192,
"step": 195
},
{
"epoch": 0.16494845360824742,
"grad_norm": 85.27603912353516,
"learning_rate": 9.904583462541357e-06,
"loss": 0.8933,
"step": 196
},
{
"epoch": 0.16579002735114665,
"grad_norm": 78.72230529785156,
"learning_rate": 9.903267108855947e-06,
"loss": 0.9672,
"step": 197
},
{
"epoch": 0.16663160109404587,
"grad_norm": 104.11307525634766,
"learning_rate": 9.90194182583218e-06,
"loss": 1.0846,
"step": 198
},
{
"epoch": 0.1674731748369451,
"grad_norm": 89.3291244506836,
"learning_rate": 9.900607615883528e-06,
"loss": 1.1313,
"step": 199
},
{
"epoch": 0.1683147485798443,
"grad_norm": 103.45892333984375,
"learning_rate": 9.89926448143972e-06,
"loss": 0.9088,
"step": 200
},
{
"epoch": 0.16915632232274352,
"grad_norm": 113.38309478759766,
"learning_rate": 9.897912424946741e-06,
"loss": 0.9909,
"step": 201
},
{
"epoch": 0.16999789606564275,
"grad_norm": 64.55664825439453,
"learning_rate": 9.896551448866817e-06,
"loss": 0.7254,
"step": 202
},
{
"epoch": 0.17083946980854198,
"grad_norm": 90.4404067993164,
"learning_rate": 9.895181555678419e-06,
"loss": 0.8409,
"step": 203
},
{
"epoch": 0.1716810435514412,
"grad_norm": 99.30708312988281,
"learning_rate": 9.893802747876264e-06,
"loss": 1.021,
"step": 204
},
{
"epoch": 0.17252261729434043,
"grad_norm": 102.0557632446289,
"learning_rate": 9.892415027971296e-06,
"loss": 0.9852,
"step": 205
},
{
"epoch": 0.17336419103723963,
"grad_norm": 88.7022476196289,
"learning_rate": 9.891018398490692e-06,
"loss": 0.9099,
"step": 206
},
{
"epoch": 0.17420576478013886,
"grad_norm": 80.32708740234375,
"learning_rate": 9.889612861977855e-06,
"loss": 0.9637,
"step": 207
},
{
"epoch": 0.17504733852303808,
"grad_norm": 84.7457504272461,
"learning_rate": 9.888198420992403e-06,
"loss": 0.8714,
"step": 208
},
{
"epoch": 0.1758889122659373,
"grad_norm": 72.03134155273438,
"learning_rate": 9.886775078110178e-06,
"loss": 1.1654,
"step": 209
},
{
"epoch": 0.17673048600883653,
"grad_norm": 135.95729064941406,
"learning_rate": 9.88534283592323e-06,
"loss": 1.0937,
"step": 210
},
{
"epoch": 0.17757205975173573,
"grad_norm": 77.91849517822266,
"learning_rate": 9.883901697039809e-06,
"loss": 1.1187,
"step": 211
},
{
"epoch": 0.17841363349463496,
"grad_norm": 66.30738830566406,
"learning_rate": 9.882451664084378e-06,
"loss": 0.8919,
"step": 212
},
{
"epoch": 0.1792552072375342,
"grad_norm": 94.25729370117188,
"learning_rate": 9.88099273969759e-06,
"loss": 1.0707,
"step": 213
},
{
"epoch": 0.1800967809804334,
"grad_norm": 131.90762329101562,
"learning_rate": 9.879524926536293e-06,
"loss": 0.8155,
"step": 214
},
{
"epoch": 0.18093835472333264,
"grad_norm": 86.80030059814453,
"learning_rate": 9.87804822727352e-06,
"loss": 1.0187,
"step": 215
},
{
"epoch": 0.18177992846623187,
"grad_norm": 100.54383087158203,
"learning_rate": 9.876562644598486e-06,
"loss": 1.0865,
"step": 216
},
{
"epoch": 0.18262150220913106,
"grad_norm": 74.09727478027344,
"learning_rate": 9.875068181216589e-06,
"loss": 0.8579,
"step": 217
},
{
"epoch": 0.1834630759520303,
"grad_norm": 81.7773208618164,
"learning_rate": 9.873564839849393e-06,
"loss": 1.0477,
"step": 218
},
{
"epoch": 0.18430464969492952,
"grad_norm": 106.73563385009766,
"learning_rate": 9.872052623234632e-06,
"loss": 0.928,
"step": 219
},
{
"epoch": 0.18514622343782874,
"grad_norm": 141.8777618408203,
"learning_rate": 9.870531534126205e-06,
"loss": 0.8244,
"step": 220
},
{
"epoch": 0.18598779718072797,
"grad_norm": 74.35907745361328,
"learning_rate": 9.869001575294165e-06,
"loss": 1.0451,
"step": 221
},
{
"epoch": 0.18682937092362717,
"grad_norm": 86.3270263671875,
"learning_rate": 9.867462749524723e-06,
"loss": 0.91,
"step": 222
},
{
"epoch": 0.1876709446665264,
"grad_norm": 103.0061264038086,
"learning_rate": 9.865915059620233e-06,
"loss": 1.1826,
"step": 223
},
{
"epoch": 0.18851251840942562,
"grad_norm": 136.33871459960938,
"learning_rate": 9.86435850839919e-06,
"loss": 1.1454,
"step": 224
},
{
"epoch": 0.18935409215232485,
"grad_norm": 125.58139038085938,
"learning_rate": 9.86279309869623e-06,
"loss": 0.961,
"step": 225
},
{
"epoch": 0.19019566589522408,
"grad_norm": 82.94192504882812,
"learning_rate": 9.861218833362122e-06,
"loss": 0.6893,
"step": 226
},
{
"epoch": 0.1910372396381233,
"grad_norm": 76.55450439453125,
"learning_rate": 9.85963571526376e-06,
"loss": 0.9668,
"step": 227
},
{
"epoch": 0.1918788133810225,
"grad_norm": 84.87679290771484,
"learning_rate": 9.858043747284159e-06,
"loss": 1.015,
"step": 228
},
{
"epoch": 0.19272038712392173,
"grad_norm": 74.23223876953125,
"learning_rate": 9.85644293232245e-06,
"loss": 0.8546,
"step": 229
},
{
"epoch": 0.19356196086682095,
"grad_norm": 91.73802185058594,
"learning_rate": 9.85483327329388e-06,
"loss": 1.0389,
"step": 230
},
{
"epoch": 0.19440353460972018,
"grad_norm": 83.0692367553711,
"learning_rate": 9.853214773129796e-06,
"loss": 1.0492,
"step": 231
},
{
"epoch": 0.1952451083526194,
"grad_norm": 86.86729431152344,
"learning_rate": 9.85158743477765e-06,
"loss": 0.9955,
"step": 232
},
{
"epoch": 0.19608668209551863,
"grad_norm": 80.8218002319336,
"learning_rate": 9.849951261200985e-06,
"loss": 1.0062,
"step": 233
},
{
"epoch": 0.19692825583841783,
"grad_norm": 68.5582275390625,
"learning_rate": 9.848306255379439e-06,
"loss": 0.8599,
"step": 234
},
{
"epoch": 0.19776982958131706,
"grad_norm": 70.14837646484375,
"learning_rate": 9.846652420308728e-06,
"loss": 0.8064,
"step": 235
},
{
"epoch": 0.19861140332421628,
"grad_norm": 67.04959106445312,
"learning_rate": 9.844989759000655e-06,
"loss": 0.8945,
"step": 236
},
{
"epoch": 0.1994529770671155,
"grad_norm": 75.14083862304688,
"learning_rate": 9.843318274483089e-06,
"loss": 0.7982,
"step": 237
},
{
"epoch": 0.20029455081001474,
"grad_norm": 60.71373748779297,
"learning_rate": 9.841637969799972e-06,
"loss": 0.8726,
"step": 238
},
{
"epoch": 0.20113612455291394,
"grad_norm": 65.94332885742188,
"learning_rate": 9.839948848011304e-06,
"loss": 0.8381,
"step": 239
},
{
"epoch": 0.20197769829581316,
"grad_norm": 78.6270523071289,
"learning_rate": 9.838250912193147e-06,
"loss": 1.0042,
"step": 240
},
{
"epoch": 0.2028192720387124,
"grad_norm": 92.58940124511719,
"learning_rate": 9.83654416543761e-06,
"loss": 0.9177,
"step": 241
},
{
"epoch": 0.20366084578161162,
"grad_norm": 98.38703155517578,
"learning_rate": 9.834828610852847e-06,
"loss": 0.9643,
"step": 242
},
{
"epoch": 0.20450241952451084,
"grad_norm": 63.133460998535156,
"learning_rate": 9.833104251563058e-06,
"loss": 0.8639,
"step": 243
},
{
"epoch": 0.20534399326741007,
"grad_norm": 80.93818664550781,
"learning_rate": 9.83137109070847e-06,
"loss": 1.1544,
"step": 244
},
{
"epoch": 0.20618556701030927,
"grad_norm": 82.82068634033203,
"learning_rate": 9.829629131445342e-06,
"loss": 0.9963,
"step": 245
},
{
"epoch": 0.2070271407532085,
"grad_norm": 76.3180923461914,
"learning_rate": 9.827878376945958e-06,
"loss": 0.8227,
"step": 246
},
{
"epoch": 0.20786871449610772,
"grad_norm": 103.49880981445312,
"learning_rate": 9.826118830398614e-06,
"loss": 0.8968,
"step": 247
},
{
"epoch": 0.20871028823900695,
"grad_norm": 73.38568878173828,
"learning_rate": 9.824350495007622e-06,
"loss": 0.9573,
"step": 248
},
{
"epoch": 0.20955186198190617,
"grad_norm": 74.37706756591797,
"learning_rate": 9.822573373993296e-06,
"loss": 0.9857,
"step": 249
},
{
"epoch": 0.2103934357248054,
"grad_norm": 67.03143310546875,
"learning_rate": 9.820787470591948e-06,
"loss": 0.9575,
"step": 250
},
{
"epoch": 0.2112350094677046,
"grad_norm": 59.41736602783203,
"learning_rate": 9.81899278805589e-06,
"loss": 0.677,
"step": 251
},
{
"epoch": 0.21207658321060383,
"grad_norm": 67.86038208007812,
"learning_rate": 9.817189329653417e-06,
"loss": 0.7332,
"step": 252
},
{
"epoch": 0.21291815695350305,
"grad_norm": 55.970970153808594,
"learning_rate": 9.815377098668806e-06,
"loss": 0.6251,
"step": 253
},
{
"epoch": 0.21375973069640228,
"grad_norm": 57.62248992919922,
"learning_rate": 9.813556098402311e-06,
"loss": 0.7727,
"step": 254
},
{
"epoch": 0.2146013044393015,
"grad_norm": 63.80696105957031,
"learning_rate": 9.811726332170153e-06,
"loss": 0.7638,
"step": 255
},
{
"epoch": 0.2154428781822007,
"grad_norm": 94.49886322021484,
"learning_rate": 9.809887803304519e-06,
"loss": 1.0082,
"step": 256
},
{
"epoch": 0.21628445192509993,
"grad_norm": 90.65179443359375,
"learning_rate": 9.808040515153556e-06,
"loss": 1.0395,
"step": 257
},
{
"epoch": 0.21712602566799916,
"grad_norm": 53.16643142700195,
"learning_rate": 9.806184471081359e-06,
"loss": 0.6617,
"step": 258
},
{
"epoch": 0.21796759941089838,
"grad_norm": 63.37334442138672,
"learning_rate": 9.804319674467969e-06,
"loss": 0.8673,
"step": 259
},
{
"epoch": 0.2188091731537976,
"grad_norm": 55.60384750366211,
"learning_rate": 9.802446128709367e-06,
"loss": 0.8789,
"step": 260
},
{
"epoch": 0.21965074689669684,
"grad_norm": 72.15032196044922,
"learning_rate": 9.800563837217465e-06,
"loss": 0.7897,
"step": 261
},
{
"epoch": 0.22049232063959603,
"grad_norm": 62.21379089355469,
"learning_rate": 9.798672803420103e-06,
"loss": 0.86,
"step": 262
},
{
"epoch": 0.22133389438249526,
"grad_norm": 94.53018951416016,
"learning_rate": 9.796773030761047e-06,
"loss": 0.9177,
"step": 263
},
{
"epoch": 0.2221754681253945,
"grad_norm": 69.56657409667969,
"learning_rate": 9.794864522699968e-06,
"loss": 0.9666,
"step": 264
},
{
"epoch": 0.22301704186829371,
"grad_norm": 61.909507751464844,
"learning_rate": 9.792947282712451e-06,
"loss": 0.8688,
"step": 265
},
{
"epoch": 0.22385861561119294,
"grad_norm": 63.15508270263672,
"learning_rate": 9.791021314289978e-06,
"loss": 0.931,
"step": 266
},
{
"epoch": 0.22470018935409214,
"grad_norm": 78.5430908203125,
"learning_rate": 9.789086620939936e-06,
"loss": 0.8776,
"step": 267
},
{
"epoch": 0.22554176309699137,
"grad_norm": 88.56510162353516,
"learning_rate": 9.78714320618559e-06,
"loss": 0.8123,
"step": 268
},
{
"epoch": 0.2263833368398906,
"grad_norm": 53.564273834228516,
"learning_rate": 9.785191073566093e-06,
"loss": 0.761,
"step": 269
},
{
"epoch": 0.22722491058278982,
"grad_norm": 64.89131927490234,
"learning_rate": 9.783230226636472e-06,
"loss": 1.0012,
"step": 270
},
{
"epoch": 0.22806648432568905,
"grad_norm": 66.71524047851562,
"learning_rate": 9.781260668967627e-06,
"loss": 0.9903,
"step": 271
},
{
"epoch": 0.22890805806858827,
"grad_norm": 83.0013198852539,
"learning_rate": 9.77928240414632e-06,
"loss": 0.9299,
"step": 272
},
{
"epoch": 0.22974963181148747,
"grad_norm": 49.404930114746094,
"learning_rate": 9.777295435775165e-06,
"loss": 0.6964,
"step": 273
},
{
"epoch": 0.2305912055543867,
"grad_norm": 61.945655822753906,
"learning_rate": 9.77529976747263e-06,
"loss": 0.7598,
"step": 274
},
{
"epoch": 0.23143277929728592,
"grad_norm": 61.8249626159668,
"learning_rate": 9.773295402873027e-06,
"loss": 0.9795,
"step": 275
},
{
"epoch": 0.23227435304018515,
"grad_norm": 54.0324592590332,
"learning_rate": 9.771282345626505e-06,
"loss": 0.9019,
"step": 276
},
{
"epoch": 0.23311592678308438,
"grad_norm": 66.27642822265625,
"learning_rate": 9.76926059939904e-06,
"loss": 0.7623,
"step": 277
},
{
"epoch": 0.2339575005259836,
"grad_norm": 70.68103790283203,
"learning_rate": 9.767230167872434e-06,
"loss": 1.1288,
"step": 278
},
{
"epoch": 0.2347990742688828,
"grad_norm": 67.17633056640625,
"learning_rate": 9.765191054744305e-06,
"loss": 1.0232,
"step": 279
},
{
"epoch": 0.23564064801178203,
"grad_norm": 46.92961883544922,
"learning_rate": 9.763143263728083e-06,
"loss": 0.8656,
"step": 280
},
{
"epoch": 0.23648222175468125,
"grad_norm": 68.37699890136719,
"learning_rate": 9.761086798552998e-06,
"loss": 1.2,
"step": 281
},
{
"epoch": 0.23732379549758048,
"grad_norm": 65.88874053955078,
"learning_rate": 9.759021662964079e-06,
"loss": 0.961,
"step": 282
},
{
"epoch": 0.2381653692404797,
"grad_norm": 70.57432556152344,
"learning_rate": 9.756947860722143e-06,
"loss": 0.9248,
"step": 283
},
{
"epoch": 0.2390069429833789,
"grad_norm": 76.35360717773438,
"learning_rate": 9.754865395603792e-06,
"loss": 0.9037,
"step": 284
},
{
"epoch": 0.23984851672627813,
"grad_norm": 64.79563903808594,
"learning_rate": 9.752774271401403e-06,
"loss": 0.7986,
"step": 285
},
{
"epoch": 0.24069009046917736,
"grad_norm": 65.64253234863281,
"learning_rate": 9.750674491923117e-06,
"loss": 0.8068,
"step": 286
},
{
"epoch": 0.24153166421207659,
"grad_norm": 62.13399887084961,
"learning_rate": 9.748566060992847e-06,
"loss": 0.6377,
"step": 287
},
{
"epoch": 0.2423732379549758,
"grad_norm": 83.9788818359375,
"learning_rate": 9.746448982450256e-06,
"loss": 0.9212,
"step": 288
},
{
"epoch": 0.24321481169787504,
"grad_norm": 57.64104461669922,
"learning_rate": 9.744323260150753e-06,
"loss": 0.9132,
"step": 289
},
{
"epoch": 0.24405638544077424,
"grad_norm": 55.18711853027344,
"learning_rate": 9.74218889796549e-06,
"loss": 0.7277,
"step": 290
},
{
"epoch": 0.24489795918367346,
"grad_norm": 53.222721099853516,
"learning_rate": 9.740045899781353e-06,
"loss": 0.7349,
"step": 291
},
{
"epoch": 0.2457395329265727,
"grad_norm": 80.6227798461914,
"learning_rate": 9.737894269500958e-06,
"loss": 0.983,
"step": 292
},
{
"epoch": 0.24658110666947192,
"grad_norm": 80.06282806396484,
"learning_rate": 9.735734011042637e-06,
"loss": 0.9006,
"step": 293
},
{
"epoch": 0.24742268041237114,
"grad_norm": 63.42185974121094,
"learning_rate": 9.733565128340434e-06,
"loss": 0.9211,
"step": 294
},
{
"epoch": 0.24826425415527034,
"grad_norm": 92.3994369506836,
"learning_rate": 9.731387625344104e-06,
"loss": 1.105,
"step": 295
},
{
"epoch": 0.24910582789816957,
"grad_norm": 99.70626068115234,
"learning_rate": 9.729201506019096e-06,
"loss": 0.9123,
"step": 296
},
{
"epoch": 0.2499474016410688,
"grad_norm": 79.17935943603516,
"learning_rate": 9.727006774346551e-06,
"loss": 1.0235,
"step": 297
},
{
"epoch": 0.250788975383968,
"grad_norm": 136.6233367919922,
"learning_rate": 9.724803434323295e-06,
"loss": 0.7044,
"step": 298
},
{
"epoch": 0.2516305491268672,
"grad_norm": 48.2202033996582,
"learning_rate": 9.722591489961829e-06,
"loss": 0.7201,
"step": 299
},
{
"epoch": 0.25247212286976645,
"grad_norm": 54.050209045410156,
"learning_rate": 9.720370945290321e-06,
"loss": 0.929,
"step": 300
},
{
"epoch": 0.2533136966126657,
"grad_norm": 48.784423828125,
"learning_rate": 9.71814180435261e-06,
"loss": 0.7137,
"step": 301
},
{
"epoch": 0.2541552703555649,
"grad_norm": 58.50812530517578,
"learning_rate": 9.715904071208182e-06,
"loss": 0.9371,
"step": 302
},
{
"epoch": 0.2549968440984641,
"grad_norm": 109.73544311523438,
"learning_rate": 9.713657749932172e-06,
"loss": 0.9712,
"step": 303
},
{
"epoch": 0.25583841784136335,
"grad_norm": 49.40225601196289,
"learning_rate": 9.711402844615355e-06,
"loss": 0.6535,
"step": 304
},
{
"epoch": 0.2566799915842626,
"grad_norm": 84.14966583251953,
"learning_rate": 9.709139359364136e-06,
"loss": 1.2019,
"step": 305
},
{
"epoch": 0.2575215653271618,
"grad_norm": 54.90718460083008,
"learning_rate": 9.706867298300552e-06,
"loss": 0.8502,
"step": 306
},
{
"epoch": 0.25836313907006103,
"grad_norm": 53.705345153808594,
"learning_rate": 9.70458666556225e-06,
"loss": 0.7891,
"step": 307
},
{
"epoch": 0.25920471281296026,
"grad_norm": 59.40304946899414,
"learning_rate": 9.702297465302488e-06,
"loss": 0.8763,
"step": 308
},
{
"epoch": 0.2600462865558595,
"grad_norm": 51.26058578491211,
"learning_rate": 9.699999701690133e-06,
"loss": 0.9267,
"step": 309
},
{
"epoch": 0.26088786029875866,
"grad_norm": 56.30524444580078,
"learning_rate": 9.697693378909636e-06,
"loss": 0.8487,
"step": 310
},
{
"epoch": 0.2617294340416579,
"grad_norm": 65.1522216796875,
"learning_rate": 9.695378501161045e-06,
"loss": 0.8303,
"step": 311
},
{
"epoch": 0.2625710077845571,
"grad_norm": 53.80173110961914,
"learning_rate": 9.693055072659981e-06,
"loss": 0.6888,
"step": 312
},
{
"epoch": 0.26341258152745634,
"grad_norm": 61.00335693359375,
"learning_rate": 9.69072309763764e-06,
"loss": 0.8475,
"step": 313
},
{
"epoch": 0.26425415527035556,
"grad_norm": 65.46568298339844,
"learning_rate": 9.688382580340781e-06,
"loss": 0.999,
"step": 314
},
{
"epoch": 0.2650957290132548,
"grad_norm": 57.85429382324219,
"learning_rate": 9.68603352503172e-06,
"loss": 0.96,
"step": 315
},
{
"epoch": 0.265937302756154,
"grad_norm": 73.73136138916016,
"learning_rate": 9.683675935988317e-06,
"loss": 1.0216,
"step": 316
},
{
"epoch": 0.26677887649905324,
"grad_norm": 69.70044708251953,
"learning_rate": 9.681309817503982e-06,
"loss": 1.1004,
"step": 317
},
{
"epoch": 0.26762045024195247,
"grad_norm": 63.45472717285156,
"learning_rate": 9.67893517388765e-06,
"loss": 0.8729,
"step": 318
},
{
"epoch": 0.2684620239848517,
"grad_norm": 57.01183319091797,
"learning_rate": 9.676552009463784e-06,
"loss": 0.9441,
"step": 319
},
{
"epoch": 0.2693035977277509,
"grad_norm": 54.7595329284668,
"learning_rate": 9.674160328572366e-06,
"loss": 0.7935,
"step": 320
},
{
"epoch": 0.2701451714706501,
"grad_norm": 45.95233154296875,
"learning_rate": 9.671760135568882e-06,
"loss": 0.6531,
"step": 321
},
{
"epoch": 0.2709867452135493,
"grad_norm": 53.747291564941406,
"learning_rate": 9.669351434824325e-06,
"loss": 0.7133,
"step": 322
},
{
"epoch": 0.27182831895644854,
"grad_norm": 53.79929733276367,
"learning_rate": 9.66693423072518e-06,
"loss": 0.7867,
"step": 323
},
{
"epoch": 0.27266989269934777,
"grad_norm": 48.851043701171875,
"learning_rate": 9.664508527673415e-06,
"loss": 0.6619,
"step": 324
},
{
"epoch": 0.273511466442247,
"grad_norm": 48.40879821777344,
"learning_rate": 9.662074330086477e-06,
"loss": 0.7443,
"step": 325
},
{
"epoch": 0.2743530401851462,
"grad_norm": 61.18216323852539,
"learning_rate": 9.659631642397283e-06,
"loss": 0.8962,
"step": 326
},
{
"epoch": 0.27519461392804545,
"grad_norm": 50.447818756103516,
"learning_rate": 9.657180469054213e-06,
"loss": 0.5483,
"step": 327
},
{
"epoch": 0.2760361876709447,
"grad_norm": 50.99171829223633,
"learning_rate": 9.654720814521094e-06,
"loss": 0.783,
"step": 328
},
{
"epoch": 0.2768777614138439,
"grad_norm": 46.67074203491211,
"learning_rate": 9.652252683277205e-06,
"loss": 0.6734,
"step": 329
},
{
"epoch": 0.27771933515674313,
"grad_norm": 58.20199966430664,
"learning_rate": 9.64977607981726e-06,
"loss": 0.8871,
"step": 330
},
{
"epoch": 0.27856090889964236,
"grad_norm": 55.42014694213867,
"learning_rate": 9.647291008651398e-06,
"loss": 1.0026,
"step": 331
},
{
"epoch": 0.2794024826425415,
"grad_norm": 57.08066177368164,
"learning_rate": 9.644797474305184e-06,
"loss": 0.9247,
"step": 332
},
{
"epoch": 0.28024405638544075,
"grad_norm": 67.04178619384766,
"learning_rate": 9.64229548131959e-06,
"loss": 0.9159,
"step": 333
},
{
"epoch": 0.28108563012834,
"grad_norm": 58.45100402832031,
"learning_rate": 9.639785034250996e-06,
"loss": 0.8391,
"step": 334
},
{
"epoch": 0.2819272038712392,
"grad_norm": 47.15868377685547,
"learning_rate": 9.637266137671177e-06,
"loss": 0.8105,
"step": 335
},
{
"epoch": 0.28276877761413843,
"grad_norm": 73.43692779541016,
"learning_rate": 9.634738796167296e-06,
"loss": 1.0237,
"step": 336
},
{
"epoch": 0.28361035135703766,
"grad_norm": 50.68719482421875,
"learning_rate": 9.63220301434189e-06,
"loss": 0.874,
"step": 337
},
{
"epoch": 0.2844519250999369,
"grad_norm": 55.89727783203125,
"learning_rate": 9.629658796812876e-06,
"loss": 0.8893,
"step": 338
},
{
"epoch": 0.2852934988428361,
"grad_norm": 60.502716064453125,
"learning_rate": 9.627106148213521e-06,
"loss": 0.9909,
"step": 339
},
{
"epoch": 0.28613507258573534,
"grad_norm": 54.480812072753906,
"learning_rate": 9.62454507319246e-06,
"loss": 0.8509,
"step": 340
},
{
"epoch": 0.28697664632863457,
"grad_norm": 50.51625442504883,
"learning_rate": 9.62197557641366e-06,
"loss": 0.7953,
"step": 341
},
{
"epoch": 0.2878182200715338,
"grad_norm": 52.29000473022461,
"learning_rate": 9.619397662556434e-06,
"loss": 0.8861,
"step": 342
},
{
"epoch": 0.28865979381443296,
"grad_norm": 88.0934829711914,
"learning_rate": 9.61681133631542e-06,
"loss": 0.8222,
"step": 343
},
{
"epoch": 0.2895013675573322,
"grad_norm": 67.22352600097656,
"learning_rate": 9.614216602400576e-06,
"loss": 0.7826,
"step": 344
},
{
"epoch": 0.2903429413002314,
"grad_norm": 48.924835205078125,
"learning_rate": 9.61161346553717e-06,
"loss": 0.8052,
"step": 345
},
{
"epoch": 0.29118451504313064,
"grad_norm": 52.71432113647461,
"learning_rate": 9.609001930465774e-06,
"loss": 0.9392,
"step": 346
},
{
"epoch": 0.29202608878602987,
"grad_norm": 47.44670104980469,
"learning_rate": 9.606382001942256e-06,
"loss": 0.9062,
"step": 347
},
{
"epoch": 0.2928676625289291,
"grad_norm": 74.1554946899414,
"learning_rate": 9.603753684737764e-06,
"loss": 0.9853,
"step": 348
},
{
"epoch": 0.2937092362718283,
"grad_norm": 63.42856216430664,
"learning_rate": 9.601116983638727e-06,
"loss": 0.8555,
"step": 349
},
{
"epoch": 0.29455081001472755,
"grad_norm": 44.9845085144043,
"learning_rate": 9.598471903446843e-06,
"loss": 0.6364,
"step": 350
},
{
"epoch": 0.2953923837576268,
"grad_norm": 56.01649856567383,
"learning_rate": 9.595818448979061e-06,
"loss": 0.9389,
"step": 351
},
{
"epoch": 0.296233957500526,
"grad_norm": 48.727935791015625,
"learning_rate": 9.593156625067592e-06,
"loss": 0.62,
"step": 352
},
{
"epoch": 0.29707553124342523,
"grad_norm": 62.60564041137695,
"learning_rate": 9.59048643655988e-06,
"loss": 1.0157,
"step": 353
},
{
"epoch": 0.2979171049863244,
"grad_norm": 54.87966537475586,
"learning_rate": 9.587807888318606e-06,
"loss": 1.0561,
"step": 354
},
{
"epoch": 0.2987586787292236,
"grad_norm": 57.15278625488281,
"learning_rate": 9.585120985221672e-06,
"loss": 1.0297,
"step": 355
},
{
"epoch": 0.29960025247212285,
"grad_norm": 63.67365264892578,
"learning_rate": 9.582425732162198e-06,
"loss": 1.0277,
"step": 356
},
{
"epoch": 0.3004418262150221,
"grad_norm": 52.32630157470703,
"learning_rate": 9.579722134048506e-06,
"loss": 0.8109,
"step": 357
},
{
"epoch": 0.3012833999579213,
"grad_norm": 45.36003875732422,
"learning_rate": 9.57701019580412e-06,
"loss": 0.7555,
"step": 358
},
{
"epoch": 0.30212497370082053,
"grad_norm": 48.94035720825195,
"learning_rate": 9.574289922367748e-06,
"loss": 0.7156,
"step": 359
},
{
"epoch": 0.30296654744371976,
"grad_norm": 50.90315246582031,
"learning_rate": 9.571561318693283e-06,
"loss": 1.0367,
"step": 360
},
{
"epoch": 0.303808121186619,
"grad_norm": 64.30524444580078,
"learning_rate": 9.568824389749781e-06,
"loss": 0.8503,
"step": 361
},
{
"epoch": 0.3046496949295182,
"grad_norm": 49.013309478759766,
"learning_rate": 9.566079140521462e-06,
"loss": 0.769,
"step": 362
},
{
"epoch": 0.30549126867241744,
"grad_norm": 50.825443267822266,
"learning_rate": 9.563325576007702e-06,
"loss": 0.8929,
"step": 363
},
{
"epoch": 0.30633284241531666,
"grad_norm": 51.26329040527344,
"learning_rate": 9.560563701223014e-06,
"loss": 0.9266,
"step": 364
},
{
"epoch": 0.3071744161582159,
"grad_norm": 48.98971176147461,
"learning_rate": 9.557793521197048e-06,
"loss": 0.9614,
"step": 365
},
{
"epoch": 0.30801598990111506,
"grad_norm": 50.82699203491211,
"learning_rate": 9.555015040974579e-06,
"loss": 0.8423,
"step": 366
},
{
"epoch": 0.3088575636440143,
"grad_norm": 48.19704818725586,
"learning_rate": 9.552228265615493e-06,
"loss": 0.8885,
"step": 367
},
{
"epoch": 0.3096991373869135,
"grad_norm": 65.29438781738281,
"learning_rate": 9.549433200194791e-06,
"loss": 1.0512,
"step": 368
},
{
"epoch": 0.31054071112981274,
"grad_norm": 61.33700942993164,
"learning_rate": 9.546629849802562e-06,
"loss": 0.854,
"step": 369
},
{
"epoch": 0.31138228487271197,
"grad_norm": 57.14670181274414,
"learning_rate": 9.543818219543988e-06,
"loss": 1.0724,
"step": 370
},
{
"epoch": 0.3122238586156112,
"grad_norm": 54.50704574584961,
"learning_rate": 9.540998314539327e-06,
"loss": 0.8919,
"step": 371
},
{
"epoch": 0.3130654323585104,
"grad_norm": 40.35006332397461,
"learning_rate": 9.53817013992391e-06,
"loss": 0.8481,
"step": 372
},
{
"epoch": 0.31390700610140965,
"grad_norm": 43.12276840209961,
"learning_rate": 9.535333700848124e-06,
"loss": 0.6807,
"step": 373
},
{
"epoch": 0.3147485798443089,
"grad_norm": 42.860713958740234,
"learning_rate": 9.532489002477406e-06,
"loss": 0.8186,
"step": 374
},
{
"epoch": 0.3155901535872081,
"grad_norm": 48.515380859375,
"learning_rate": 9.529636049992235e-06,
"loss": 0.9689,
"step": 375
},
{
"epoch": 0.3164317273301073,
"grad_norm": 109.04833984375,
"learning_rate": 9.526774848588122e-06,
"loss": 0.9041,
"step": 376
},
{
"epoch": 0.3172733010730065,
"grad_norm": 49.20893478393555,
"learning_rate": 9.523905403475604e-06,
"loss": 0.8922,
"step": 377
},
{
"epoch": 0.3181148748159057,
"grad_norm": 38.91448974609375,
"learning_rate": 9.521027719880223e-06,
"loss": 0.7721,
"step": 378
},
{
"epoch": 0.31895644855880495,
"grad_norm": 48.46686553955078,
"learning_rate": 9.518141803042528e-06,
"loss": 0.8912,
"step": 379
},
{
"epoch": 0.3197980223017042,
"grad_norm": 49.667720794677734,
"learning_rate": 9.51524765821806e-06,
"loss": 0.742,
"step": 380
},
{
"epoch": 0.3206395960446034,
"grad_norm": 40.567867279052734,
"learning_rate": 9.51234529067735e-06,
"loss": 0.7076,
"step": 381
},
{
"epoch": 0.32148116978750263,
"grad_norm": 49.25556945800781,
"learning_rate": 9.509434705705895e-06,
"loss": 0.7404,
"step": 382
},
{
"epoch": 0.32232274353040186,
"grad_norm": 45.33151626586914,
"learning_rate": 9.506515908604162e-06,
"loss": 0.7512,
"step": 383
},
{
"epoch": 0.3231643172733011,
"grad_norm": 51.83771896362305,
"learning_rate": 9.50358890468757e-06,
"loss": 1.0436,
"step": 384
},
{
"epoch": 0.3240058910162003,
"grad_norm": 49.51242446899414,
"learning_rate": 9.500653699286488e-06,
"loss": 0.6772,
"step": 385
},
{
"epoch": 0.32484746475909954,
"grad_norm": 59.32050323486328,
"learning_rate": 9.497710297746214e-06,
"loss": 1.0014,
"step": 386
},
{
"epoch": 0.32568903850199876,
"grad_norm": 42.394344329833984,
"learning_rate": 9.494758705426978e-06,
"loss": 0.5142,
"step": 387
},
{
"epoch": 0.32653061224489793,
"grad_norm": 46.40264892578125,
"learning_rate": 9.491798927703922e-06,
"loss": 0.7008,
"step": 388
},
{
"epoch": 0.32737218598779716,
"grad_norm": 51.24726867675781,
"learning_rate": 9.4888309699671e-06,
"loss": 0.9703,
"step": 389
},
{
"epoch": 0.3282137597306964,
"grad_norm": 48.097652435302734,
"learning_rate": 9.485854837621455e-06,
"loss": 0.8275,
"step": 390
},
{
"epoch": 0.3290553334735956,
"grad_norm": 57.137718200683594,
"learning_rate": 9.482870536086824e-06,
"loss": 1.0279,
"step": 391
},
{
"epoch": 0.32989690721649484,
"grad_norm": 52.86214065551758,
"learning_rate": 9.479878070797916e-06,
"loss": 0.742,
"step": 392
},
{
"epoch": 0.33073848095939407,
"grad_norm": 46.32011413574219,
"learning_rate": 9.476877447204308e-06,
"loss": 0.7686,
"step": 393
},
{
"epoch": 0.3315800547022933,
"grad_norm": 42.45808029174805,
"learning_rate": 9.473868670770436e-06,
"loss": 0.6933,
"step": 394
},
{
"epoch": 0.3324216284451925,
"grad_norm": 42.44548416137695,
"learning_rate": 9.470851746975582e-06,
"loss": 0.8593,
"step": 395
},
{
"epoch": 0.33326320218809175,
"grad_norm": 53.4603157043457,
"learning_rate": 9.467826681313865e-06,
"loss": 0.9433,
"step": 396
},
{
"epoch": 0.33410477593099097,
"grad_norm": 44.40301513671875,
"learning_rate": 9.464793479294232e-06,
"loss": 0.6444,
"step": 397
},
{
"epoch": 0.3349463496738902,
"grad_norm": 40.066165924072266,
"learning_rate": 9.461752146440443e-06,
"loss": 0.8588,
"step": 398
},
{
"epoch": 0.33578792341678937,
"grad_norm": 39.62493896484375,
"learning_rate": 9.458702688291072e-06,
"loss": 0.7671,
"step": 399
},
{
"epoch": 0.3366294971596886,
"grad_norm": 39.067169189453125,
"learning_rate": 9.455645110399485e-06,
"loss": 0.6831,
"step": 400
},
{
"epoch": 0.3374710709025878,
"grad_norm": 42.04105758666992,
"learning_rate": 9.452579418333839e-06,
"loss": 0.8142,
"step": 401
},
{
"epoch": 0.33831264464548705,
"grad_norm": 53.11723709106445,
"learning_rate": 9.449505617677059e-06,
"loss": 1.0433,
"step": 402
},
{
"epoch": 0.3391542183883863,
"grad_norm": 47.59870147705078,
"learning_rate": 9.446423714026846e-06,
"loss": 0.8461,
"step": 403
},
{
"epoch": 0.3399957921312855,
"grad_norm": 41.67499923706055,
"learning_rate": 9.443333712995655e-06,
"loss": 0.7845,
"step": 404
},
{
"epoch": 0.34083736587418473,
"grad_norm": 52.8908576965332,
"learning_rate": 9.440235620210682e-06,
"loss": 0.9642,
"step": 405
},
{
"epoch": 0.34167893961708395,
"grad_norm": 44.62261962890625,
"learning_rate": 9.437129441313868e-06,
"loss": 0.7617,
"step": 406
},
{
"epoch": 0.3425205133599832,
"grad_norm": 45.07094955444336,
"learning_rate": 9.434015181961873e-06,
"loss": 0.7626,
"step": 407
},
{
"epoch": 0.3433620871028824,
"grad_norm": 54.96581268310547,
"learning_rate": 9.430892847826071e-06,
"loss": 0.9612,
"step": 408
},
{
"epoch": 0.34420366084578163,
"grad_norm": 50.503761291503906,
"learning_rate": 9.427762444592548e-06,
"loss": 0.8697,
"step": 409
},
{
"epoch": 0.34504523458868086,
"grad_norm": 55.580806732177734,
"learning_rate": 9.424623977962075e-06,
"loss": 0.7447,
"step": 410
},
{
"epoch": 0.34588680833158003,
"grad_norm": 51.2252311706543,
"learning_rate": 9.421477453650118e-06,
"loss": 0.8092,
"step": 411
},
{
"epoch": 0.34672838207447926,
"grad_norm": 55.371795654296875,
"learning_rate": 9.41832287738681e-06,
"loss": 0.7196,
"step": 412
},
{
"epoch": 0.3475699558173785,
"grad_norm": 58.95374298095703,
"learning_rate": 9.415160254916949e-06,
"loss": 1.0793,
"step": 413
},
{
"epoch": 0.3484115295602777,
"grad_norm": 59.95231628417969,
"learning_rate": 9.411989591999982e-06,
"loss": 1.1717,
"step": 414
},
{
"epoch": 0.34925310330317694,
"grad_norm": 55.32718276977539,
"learning_rate": 9.408810894410008e-06,
"loss": 0.8643,
"step": 415
},
{
"epoch": 0.35009467704607616,
"grad_norm": 39.598426818847656,
"learning_rate": 9.405624167935751e-06,
"loss": 0.8807,
"step": 416
},
{
"epoch": 0.3509362507889754,
"grad_norm": 33.940364837646484,
"learning_rate": 9.402429418380555e-06,
"loss": 0.5736,
"step": 417
},
{
"epoch": 0.3517778245318746,
"grad_norm": 36.997406005859375,
"learning_rate": 9.399226651562379e-06,
"loss": 0.7671,
"step": 418
},
{
"epoch": 0.35261939827477384,
"grad_norm": 46.59873962402344,
"learning_rate": 9.396015873313781e-06,
"loss": 0.8599,
"step": 419
},
{
"epoch": 0.35346097201767307,
"grad_norm": 52.79997634887695,
"learning_rate": 9.39279708948191e-06,
"loss": 0.8777,
"step": 420
},
{
"epoch": 0.3543025457605723,
"grad_norm": 38.179168701171875,
"learning_rate": 9.389570305928489e-06,
"loss": 0.8467,
"step": 421
},
{
"epoch": 0.35514411950347147,
"grad_norm": 45.605918884277344,
"learning_rate": 9.386335528529815e-06,
"loss": 0.904,
"step": 422
},
{
"epoch": 0.3559856932463707,
"grad_norm": 40.07881546020508,
"learning_rate": 9.38309276317674e-06,
"loss": 0.8852,
"step": 423
},
{
"epoch": 0.3568272669892699,
"grad_norm": 36.18572998046875,
"learning_rate": 9.37984201577466e-06,
"loss": 0.7763,
"step": 424
},
{
"epoch": 0.35766884073216915,
"grad_norm": 38.40799331665039,
"learning_rate": 9.376583292243515e-06,
"loss": 0.7169,
"step": 425
},
{
"epoch": 0.3585104144750684,
"grad_norm": 58.750606536865234,
"learning_rate": 9.373316598517762e-06,
"loss": 0.901,
"step": 426
},
{
"epoch": 0.3593519882179676,
"grad_norm": 44.45072937011719,
"learning_rate": 9.37004194054638e-06,
"loss": 0.835,
"step": 427
},
{
"epoch": 0.3601935619608668,
"grad_norm": 67.9157943725586,
"learning_rate": 9.366759324292843e-06,
"loss": 0.9723,
"step": 428
},
{
"epoch": 0.36103513570376605,
"grad_norm": 46.094539642333984,
"learning_rate": 9.363468755735123e-06,
"loss": 0.8572,
"step": 429
},
{
"epoch": 0.3618767094466653,
"grad_norm": 59.279029846191406,
"learning_rate": 9.360170240865677e-06,
"loss": 0.937,
"step": 430
},
{
"epoch": 0.3627182831895645,
"grad_norm": 40.15589904785156,
"learning_rate": 9.356863785691428e-06,
"loss": 0.7079,
"step": 431
},
{
"epoch": 0.36355985693246373,
"grad_norm": 39.20061111450195,
"learning_rate": 9.35354939623376e-06,
"loss": 0.7623,
"step": 432
},
{
"epoch": 0.3644014306753629,
"grad_norm": 42.307823181152344,
"learning_rate": 9.350227078528505e-06,
"loss": 0.8366,
"step": 433
},
{
"epoch": 0.36524300441826213,
"grad_norm": 45.434696197509766,
"learning_rate": 9.346896838625938e-06,
"loss": 0.9978,
"step": 434
},
{
"epoch": 0.36608457816116136,
"grad_norm": 38.71270751953125,
"learning_rate": 9.343558682590757e-06,
"loss": 0.7336,
"step": 435
},
{
"epoch": 0.3669261519040606,
"grad_norm": 56.99869918823242,
"learning_rate": 9.340212616502075e-06,
"loss": 0.8806,
"step": 436
},
{
"epoch": 0.3677677256469598,
"grad_norm": 48.77510452270508,
"learning_rate": 9.336858646453415e-06,
"loss": 0.8885,
"step": 437
},
{
"epoch": 0.36860929938985904,
"grad_norm": 39.787052154541016,
"learning_rate": 9.333496778552689e-06,
"loss": 0.5919,
"step": 438
},
{
"epoch": 0.36945087313275826,
"grad_norm": 47.59059143066406,
"learning_rate": 9.330127018922195e-06,
"loss": 0.9144,
"step": 439
},
{
"epoch": 0.3702924468756575,
"grad_norm": 40.901973724365234,
"learning_rate": 9.3267493736986e-06,
"loss": 0.9209,
"step": 440
},
{
"epoch": 0.3711340206185567,
"grad_norm": 44.961421966552734,
"learning_rate": 9.323363849032933e-06,
"loss": 0.814,
"step": 441
},
{
"epoch": 0.37197559436145594,
"grad_norm": 44.27064514160156,
"learning_rate": 9.319970451090573e-06,
"loss": 0.7334,
"step": 442
},
{
"epoch": 0.37281716810435517,
"grad_norm": 42.544124603271484,
"learning_rate": 9.316569186051234e-06,
"loss": 0.8683,
"step": 443
},
{
"epoch": 0.37365874184725434,
"grad_norm": 40.91048049926758,
"learning_rate": 9.313160060108961e-06,
"loss": 0.6704,
"step": 444
},
{
"epoch": 0.37450031559015357,
"grad_norm": 38.124969482421875,
"learning_rate": 9.309743079472112e-06,
"loss": 0.854,
"step": 445
},
{
"epoch": 0.3753418893330528,
"grad_norm": 45.910465240478516,
"learning_rate": 9.30631825036335e-06,
"loss": 0.7315,
"step": 446
},
{
"epoch": 0.376183463075952,
"grad_norm": 52.38642883300781,
"learning_rate": 9.302885579019626e-06,
"loss": 0.8705,
"step": 447
},
{
"epoch": 0.37702503681885124,
"grad_norm": 53.45127868652344,
"learning_rate": 9.299445071692183e-06,
"loss": 1.0807,
"step": 448
},
{
"epoch": 0.37786661056175047,
"grad_norm": 33.41383361816406,
"learning_rate": 9.295996734646525e-06,
"loss": 0.8253,
"step": 449
},
{
"epoch": 0.3787081843046497,
"grad_norm": 53.72993469238281,
"learning_rate": 9.292540574162416e-06,
"loss": 1.0593,
"step": 450
},
{
"epoch": 0.3795497580475489,
"grad_norm": 38.741783142089844,
"learning_rate": 9.289076596533873e-06,
"loss": 0.8046,
"step": 451
},
{
"epoch": 0.38039133179044815,
"grad_norm": 36.7025146484375,
"learning_rate": 9.28560480806914e-06,
"loss": 0.6904,
"step": 452
},
{
"epoch": 0.3812329055333474,
"grad_norm": 39.85129165649414,
"learning_rate": 9.282125215090694e-06,
"loss": 0.8253,
"step": 453
},
{
"epoch": 0.3820744792762466,
"grad_norm": 46.72287368774414,
"learning_rate": 9.278637823935217e-06,
"loss": 0.7278,
"step": 454
},
{
"epoch": 0.38291605301914583,
"grad_norm": 66.36459350585938,
"learning_rate": 9.275142640953599e-06,
"loss": 1.2174,
"step": 455
},
{
"epoch": 0.383757626762045,
"grad_norm": 33.20828628540039,
"learning_rate": 9.271639672510917e-06,
"loss": 0.6525,
"step": 456
},
{
"epoch": 0.3845992005049442,
"grad_norm": 54.37870788574219,
"learning_rate": 9.268128924986423e-06,
"loss": 1.007,
"step": 457
},
{
"epoch": 0.38544077424784345,
"grad_norm": 39.28815841674805,
"learning_rate": 9.264610404773542e-06,
"loss": 0.6188,
"step": 458
},
{
"epoch": 0.3862823479907427,
"grad_norm": 38.080047607421875,
"learning_rate": 9.261084118279846e-06,
"loss": 0.6616,
"step": 459
},
{
"epoch": 0.3871239217336419,
"grad_norm": 72.71385955810547,
"learning_rate": 9.257550071927059e-06,
"loss": 1.0173,
"step": 460
},
{
"epoch": 0.38796549547654113,
"grad_norm": 36.798587799072266,
"learning_rate": 9.254008272151029e-06,
"loss": 0.82,
"step": 461
},
{
"epoch": 0.38880706921944036,
"grad_norm": 35.71184158325195,
"learning_rate": 9.250458725401726e-06,
"loss": 0.8262,
"step": 462
},
{
"epoch": 0.3896486429623396,
"grad_norm": 28.581558227539062,
"learning_rate": 9.24690143814323e-06,
"loss": 0.5622,
"step": 463
},
{
"epoch": 0.3904902167052388,
"grad_norm": 40.32032012939453,
"learning_rate": 9.243336416853714e-06,
"loss": 0.995,
"step": 464
},
{
"epoch": 0.39133179044813804,
"grad_norm": 39.23225784301758,
"learning_rate": 9.23976366802544e-06,
"loss": 0.8746,
"step": 465
},
{
"epoch": 0.39217336419103727,
"grad_norm": 41.329246520996094,
"learning_rate": 9.236183198164739e-06,
"loss": 0.8849,
"step": 466
},
{
"epoch": 0.39301493793393644,
"grad_norm": 39.20853042602539,
"learning_rate": 9.232595013792004e-06,
"loss": 0.6925,
"step": 467
},
{
"epoch": 0.39385651167683566,
"grad_norm": 47.32665252685547,
"learning_rate": 9.228999121441673e-06,
"loss": 0.9999,
"step": 468
},
{
"epoch": 0.3946980854197349,
"grad_norm": 42.45283889770508,
"learning_rate": 9.225395527662228e-06,
"loss": 0.838,
"step": 469
},
{
"epoch": 0.3955396591626341,
"grad_norm": 36.82233810424805,
"learning_rate": 9.221784239016174e-06,
"loss": 0.8193,
"step": 470
},
{
"epoch": 0.39638123290553334,
"grad_norm": 51.216712951660156,
"learning_rate": 9.218165262080024e-06,
"loss": 1.0036,
"step": 471
},
{
"epoch": 0.39722280664843257,
"grad_norm": 31.790563583374023,
"learning_rate": 9.214538603444297e-06,
"loss": 0.7464,
"step": 472
},
{
"epoch": 0.3980643803913318,
"grad_norm": 43.2595100402832,
"learning_rate": 9.210904269713503e-06,
"loss": 0.785,
"step": 473
},
{
"epoch": 0.398905954134231,
"grad_norm": 38.354549407958984,
"learning_rate": 9.207262267506122e-06,
"loss": 0.9315,
"step": 474
},
{
"epoch": 0.39974752787713025,
"grad_norm": 64.44866943359375,
"learning_rate": 9.203612603454605e-06,
"loss": 0.86,
"step": 475
},
{
"epoch": 0.4005891016200295,
"grad_norm": 48.267555236816406,
"learning_rate": 9.199955284205355e-06,
"loss": 0.8267,
"step": 476
},
{
"epoch": 0.4014306753629287,
"grad_norm": 44.975738525390625,
"learning_rate": 9.196290316418712e-06,
"loss": 0.9038,
"step": 477
},
{
"epoch": 0.4022722491058279,
"grad_norm": 54.67453384399414,
"learning_rate": 9.192617706768952e-06,
"loss": 0.8733,
"step": 478
},
{
"epoch": 0.4031138228487271,
"grad_norm": 52.223060607910156,
"learning_rate": 9.18893746194426e-06,
"loss": 1.017,
"step": 479
},
{
"epoch": 0.4039553965916263,
"grad_norm": 41.82644271850586,
"learning_rate": 9.185249588646731e-06,
"loss": 0.8636,
"step": 480
},
{
"epoch": 0.40479697033452555,
"grad_norm": 50.87776565551758,
"learning_rate": 9.181554093592351e-06,
"loss": 0.9287,
"step": 481
},
{
"epoch": 0.4056385440774248,
"grad_norm": 40.91377258300781,
"learning_rate": 9.17785098351098e-06,
"loss": 0.6688,
"step": 482
},
{
"epoch": 0.406480117820324,
"grad_norm": 49.19743728637695,
"learning_rate": 9.174140265146356e-06,
"loss": 0.7805,
"step": 483
},
{
"epoch": 0.40732169156322323,
"grad_norm": 59.54707336425781,
"learning_rate": 9.170421945256065e-06,
"loss": 1.2094,
"step": 484
},
{
"epoch": 0.40816326530612246,
"grad_norm": 41.340782165527344,
"learning_rate": 9.16669603061154e-06,
"loss": 0.9054,
"step": 485
},
{
"epoch": 0.4090048390490217,
"grad_norm": 40.847469329833984,
"learning_rate": 9.162962527998038e-06,
"loss": 0.899,
"step": 486
},
{
"epoch": 0.4098464127919209,
"grad_norm": 51.336029052734375,
"learning_rate": 9.159221444214645e-06,
"loss": 0.8226,
"step": 487
},
{
"epoch": 0.41068798653482014,
"grad_norm": 44.75333786010742,
"learning_rate": 9.155472786074247e-06,
"loss": 0.6851,
"step": 488
},
{
"epoch": 0.4115295602777193,
"grad_norm": 50.3229866027832,
"learning_rate": 9.151716560403521e-06,
"loss": 0.9122,
"step": 489
},
{
"epoch": 0.41237113402061853,
"grad_norm": 45.247196197509766,
"learning_rate": 9.14795277404293e-06,
"loss": 0.8831,
"step": 490
},
{
"epoch": 0.41321270776351776,
"grad_norm": 43.83192825317383,
"learning_rate": 9.144181433846707e-06,
"loss": 0.8944,
"step": 491
},
{
"epoch": 0.414054281506417,
"grad_norm": 41.78791809082031,
"learning_rate": 9.140402546682837e-06,
"loss": 0.804,
"step": 492
},
{
"epoch": 0.4148958552493162,
"grad_norm": 35.123008728027344,
"learning_rate": 9.136616119433047e-06,
"loss": 0.6715,
"step": 493
},
{
"epoch": 0.41573742899221544,
"grad_norm": 36.4639892578125,
"learning_rate": 9.132822158992802e-06,
"loss": 0.9032,
"step": 494
},
{
"epoch": 0.41657900273511467,
"grad_norm": 33.941078186035156,
"learning_rate": 9.129020672271283e-06,
"loss": 0.7585,
"step": 495
},
{
"epoch": 0.4174205764780139,
"grad_norm": 43.93740463256836,
"learning_rate": 9.125211666191374e-06,
"loss": 0.8093,
"step": 496
},
{
"epoch": 0.4182621502209131,
"grad_norm": 38.40201187133789,
"learning_rate": 9.121395147689658e-06,
"loss": 0.7478,
"step": 497
},
{
"epoch": 0.41910372396381235,
"grad_norm": 36.99344253540039,
"learning_rate": 9.117571123716392e-06,
"loss": 0.614,
"step": 498
},
{
"epoch": 0.4199452977067116,
"grad_norm": 42.064762115478516,
"learning_rate": 9.113739601235508e-06,
"loss": 0.7893,
"step": 499
},
{
"epoch": 0.4207868714496108,
"grad_norm": 41.879425048828125,
"learning_rate": 9.109900587224587e-06,
"loss": 0.8887,
"step": 500
}
],
"logging_steps": 1,
"max_steps": 2378,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 2.7603496702462464e+16,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}