wangrongsheng
commit from root
638411d
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.6987542468856174,
"global_step": 6000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.9999752766303955e-05,
"loss": 2.7224,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 4.999901107010577e-05,
"loss": 2.6131,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 4.999777492607524e-05,
"loss": 2.5716,
"step": 30
},
{
"epoch": 0.01,
"learning_rate": 4.9996044358661676e-05,
"loss": 2.5126,
"step": 40
},
{
"epoch": 0.01,
"learning_rate": 4.9993819402093446e-05,
"loss": 2.4635,
"step": 50
},
{
"epoch": 0.02,
"learning_rate": 4.999110010037728e-05,
"loss": 2.4298,
"step": 60
},
{
"epoch": 0.02,
"learning_rate": 4.998788650729743e-05,
"loss": 2.4118,
"step": 70
},
{
"epoch": 0.02,
"learning_rate": 4.998417868641457e-05,
"loss": 2.416,
"step": 80
},
{
"epoch": 0.03,
"learning_rate": 4.9979976711064556e-05,
"loss": 2.4218,
"step": 90
},
{
"epoch": 0.03,
"learning_rate": 4.9975280664357e-05,
"loss": 2.4266,
"step": 100
},
{
"epoch": 0.03,
"learning_rate": 4.997009063917356e-05,
"loss": 2.4419,
"step": 110
},
{
"epoch": 0.03,
"learning_rate": 4.996440673816617e-05,
"loss": 2.3756,
"step": 120
},
{
"epoch": 0.04,
"learning_rate": 4.995822907375498e-05,
"loss": 2.4077,
"step": 130
},
{
"epoch": 0.04,
"learning_rate": 4.9951557768126136e-05,
"loss": 2.3502,
"step": 140
},
{
"epoch": 0.04,
"learning_rate": 4.9944392953229365e-05,
"loss": 2.3754,
"step": 150
},
{
"epoch": 0.05,
"learning_rate": 4.993673477077535e-05,
"loss": 2.3069,
"step": 160
},
{
"epoch": 0.05,
"learning_rate": 4.9928583372232954e-05,
"loss": 2.315,
"step": 170
},
{
"epoch": 0.05,
"learning_rate": 4.991993891882622e-05,
"loss": 2.3428,
"step": 180
},
{
"epoch": 0.05,
"learning_rate": 4.991080158153115e-05,
"loss": 2.3674,
"step": 190
},
{
"epoch": 0.06,
"learning_rate": 4.990117154107236e-05,
"loss": 2.3312,
"step": 200
},
{
"epoch": 0.06,
"learning_rate": 4.989104898791949e-05,
"loss": 2.3763,
"step": 210
},
{
"epoch": 0.06,
"learning_rate": 4.988043412228343e-05,
"loss": 2.3545,
"step": 220
},
{
"epoch": 0.07,
"learning_rate": 4.986932715411239e-05,
"loss": 2.3477,
"step": 230
},
{
"epoch": 0.07,
"learning_rate": 4.985772830308771e-05,
"loss": 2.338,
"step": 240
},
{
"epoch": 0.07,
"learning_rate": 4.9845637798619536e-05,
"loss": 2.3522,
"step": 250
},
{
"epoch": 0.07,
"learning_rate": 4.983305587984227e-05,
"loss": 2.3603,
"step": 260
},
{
"epoch": 0.08,
"learning_rate": 4.981998279560986e-05,
"loss": 2.3292,
"step": 270
},
{
"epoch": 0.08,
"learning_rate": 4.980641880449086e-05,
"loss": 2.3856,
"step": 280
},
{
"epoch": 0.08,
"learning_rate": 4.9792364174763323e-05,
"loss": 2.3062,
"step": 290
},
{
"epoch": 0.08,
"learning_rate": 4.977781918440949e-05,
"loss": 2.3289,
"step": 300
},
{
"epoch": 0.09,
"learning_rate": 4.97627841211103e-05,
"loss": 2.3062,
"step": 310
},
{
"epoch": 0.09,
"learning_rate": 4.97472592822397e-05,
"loss": 2.2828,
"step": 320
},
{
"epoch": 0.09,
"learning_rate": 4.9731244974858746e-05,
"loss": 2.3063,
"step": 330
},
{
"epoch": 0.1,
"learning_rate": 4.971474151570955e-05,
"loss": 2.3095,
"step": 340
},
{
"epoch": 0.1,
"learning_rate": 4.9697749231209025e-05,
"loss": 2.3358,
"step": 350
},
{
"epoch": 0.1,
"learning_rate": 4.968026845744237e-05,
"loss": 2.2997,
"step": 360
},
{
"epoch": 0.1,
"learning_rate": 4.96622995401565e-05,
"loss": 2.2737,
"step": 370
},
{
"epoch": 0.11,
"learning_rate": 4.9643842834753174e-05,
"loss": 2.2845,
"step": 380
},
{
"epoch": 0.11,
"learning_rate": 4.962489870628193e-05,
"loss": 2.279,
"step": 390
},
{
"epoch": 0.11,
"learning_rate": 4.960546752943294e-05,
"loss": 2.3093,
"step": 400
},
{
"epoch": 0.12,
"learning_rate": 4.958554968852952e-05,
"loss": 2.2509,
"step": 410
},
{
"epoch": 0.12,
"learning_rate": 4.95651455775206e-05,
"loss": 2.2611,
"step": 420
},
{
"epoch": 0.12,
"learning_rate": 4.954425559997287e-05,
"loss": 2.3046,
"step": 430
},
{
"epoch": 0.12,
"learning_rate": 4.952288016906284e-05,
"loss": 2.2862,
"step": 440
},
{
"epoch": 0.13,
"learning_rate": 4.9501019707568665e-05,
"loss": 2.2552,
"step": 450
},
{
"epoch": 0.13,
"learning_rate": 4.947867464786174e-05,
"loss": 2.2684,
"step": 460
},
{
"epoch": 0.13,
"learning_rate": 4.945584543189823e-05,
"loss": 2.3095,
"step": 470
},
{
"epoch": 0.14,
"learning_rate": 4.943253251121022e-05,
"loss": 2.2867,
"step": 480
},
{
"epoch": 0.14,
"learning_rate": 4.9408736346896894e-05,
"loss": 2.2652,
"step": 490
},
{
"epoch": 0.14,
"learning_rate": 4.938445740961534e-05,
"loss": 2.308,
"step": 500
},
{
"epoch": 0.14,
"learning_rate": 4.935969617957126e-05,
"loss": 2.2597,
"step": 510
},
{
"epoch": 0.15,
"learning_rate": 4.93344531465095e-05,
"loss": 2.276,
"step": 520
},
{
"epoch": 0.15,
"learning_rate": 4.930872880970432e-05,
"loss": 2.2597,
"step": 530
},
{
"epoch": 0.15,
"learning_rate": 4.9282523677949556e-05,
"loss": 2.307,
"step": 540
},
{
"epoch": 0.16,
"learning_rate": 4.9255838269548524e-05,
"loss": 2.2805,
"step": 550
},
{
"epoch": 0.16,
"learning_rate": 4.9228673112303814e-05,
"loss": 2.2425,
"step": 560
},
{
"epoch": 0.16,
"learning_rate": 4.9201028743506794e-05,
"loss": 2.2728,
"step": 570
},
{
"epoch": 0.16,
"learning_rate": 4.917290570992702e-05,
"loss": 2.2632,
"step": 580
},
{
"epoch": 0.17,
"learning_rate": 4.9144304567801415e-05,
"loss": 2.2604,
"step": 590
},
{
"epoch": 0.17,
"learning_rate": 4.911522588282327e-05,
"loss": 2.2868,
"step": 600
},
{
"epoch": 0.17,
"learning_rate": 4.908567023013104e-05,
"loss": 2.2433,
"step": 610
},
{
"epoch": 0.18,
"learning_rate": 4.9055638194296994e-05,
"loss": 2.2418,
"step": 620
},
{
"epoch": 0.18,
"learning_rate": 4.902513036931562e-05,
"loss": 2.2624,
"step": 630
},
{
"epoch": 0.18,
"learning_rate": 4.89941473585919e-05,
"loss": 2.2976,
"step": 640
},
{
"epoch": 0.18,
"learning_rate": 4.89626897749294e-05,
"loss": 2.2622,
"step": 650
},
{
"epoch": 0.19,
"learning_rate": 4.893075824051807e-05,
"loss": 2.2527,
"step": 660
},
{
"epoch": 0.19,
"learning_rate": 4.889835338692201e-05,
"loss": 2.2508,
"step": 670
},
{
"epoch": 0.19,
"learning_rate": 4.886547585506698e-05,
"loss": 2.2842,
"step": 680
},
{
"epoch": 0.2,
"learning_rate": 4.8832126295227656e-05,
"loss": 2.2229,
"step": 690
},
{
"epoch": 0.2,
"learning_rate": 4.879830536701485e-05,
"loss": 2.2499,
"step": 700
},
{
"epoch": 0.2,
"learning_rate": 4.876401373936239e-05,
"loss": 2.257,
"step": 710
},
{
"epoch": 0.2,
"learning_rate": 4.8729252090513964e-05,
"loss": 2.2497,
"step": 720
},
{
"epoch": 0.21,
"learning_rate": 4.869402110800963e-05,
"loss": 2.2429,
"step": 730
},
{
"epoch": 0.21,
"learning_rate": 4.865832148867228e-05,
"loss": 2.2452,
"step": 740
},
{
"epoch": 0.21,
"learning_rate": 4.8622153938593816e-05,
"loss": 2.2579,
"step": 750
},
{
"epoch": 0.22,
"learning_rate": 4.8585519173121205e-05,
"loss": 2.264,
"step": 760
},
{
"epoch": 0.22,
"learning_rate": 4.8548417916842324e-05,
"loss": 2.2591,
"step": 770
},
{
"epoch": 0.22,
"learning_rate": 4.851085090357163e-05,
"loss": 2.2939,
"step": 780
},
{
"epoch": 0.22,
"learning_rate": 4.847281887633565e-05,
"loss": 2.231,
"step": 790
},
{
"epoch": 0.23,
"learning_rate": 4.843432258735827e-05,
"loss": 2.2329,
"step": 800
},
{
"epoch": 0.23,
"learning_rate": 4.839536279804589e-05,
"loss": 2.2781,
"step": 810
},
{
"epoch": 0.23,
"learning_rate": 4.835594027897231e-05,
"loss": 2.2041,
"step": 820
},
{
"epoch": 0.23,
"learning_rate": 4.831605580986355e-05,
"loss": 2.2818,
"step": 830
},
{
"epoch": 0.24,
"learning_rate": 4.8275710179582366e-05,
"loss": 2.2479,
"step": 840
},
{
"epoch": 0.24,
"learning_rate": 4.823490418611273e-05,
"loss": 2.2136,
"step": 850
},
{
"epoch": 0.24,
"learning_rate": 4.819363863654395e-05,
"loss": 2.2613,
"step": 860
},
{
"epoch": 0.25,
"learning_rate": 4.8151914347054776e-05,
"loss": 2.205,
"step": 870
},
{
"epoch": 0.25,
"learning_rate": 4.810973214289724e-05,
"loss": 2.2389,
"step": 880
},
{
"epoch": 0.25,
"learning_rate": 4.806709285838031e-05,
"loss": 2.2404,
"step": 890
},
{
"epoch": 0.25,
"learning_rate": 4.8023997336853434e-05,
"loss": 2.2377,
"step": 900
},
{
"epoch": 0.26,
"learning_rate": 4.7980446430689804e-05,
"loss": 2.2662,
"step": 910
},
{
"epoch": 0.26,
"learning_rate": 4.793644100126954e-05,
"loss": 2.2298,
"step": 920
},
{
"epoch": 0.26,
"learning_rate": 4.7891981918962656e-05,
"loss": 2.2265,
"step": 930
},
{
"epoch": 0.27,
"learning_rate": 4.784707006311179e-05,
"loss": 2.2413,
"step": 940
},
{
"epoch": 0.27,
"learning_rate": 4.780170632201487e-05,
"loss": 2.2412,
"step": 950
},
{
"epoch": 0.27,
"learning_rate": 4.7755891592907556e-05,
"loss": 2.2355,
"step": 960
},
{
"epoch": 0.27,
"learning_rate": 4.7709626781945397e-05,
"loss": 2.1954,
"step": 970
},
{
"epoch": 0.28,
"learning_rate": 4.766291280418603e-05,
"loss": 2.232,
"step": 980
},
{
"epoch": 0.28,
"learning_rate": 4.761575058357101e-05,
"loss": 2.2339,
"step": 990
},
{
"epoch": 0.28,
"learning_rate": 4.756814105290753e-05,
"loss": 2.2341,
"step": 1000
},
{
"epoch": 0.29,
"learning_rate": 4.7520085153850014e-05,
"loss": 2.2251,
"step": 1010
},
{
"epoch": 0.29,
"learning_rate": 4.7471583836881476e-05,
"loss": 2.24,
"step": 1020
},
{
"epoch": 0.29,
"learning_rate": 4.742263806129469e-05,
"loss": 2.2173,
"step": 1030
},
{
"epoch": 0.29,
"learning_rate": 4.737324879517326e-05,
"loss": 2.2581,
"step": 1040
},
{
"epoch": 0.3,
"learning_rate": 4.732341701537245e-05,
"loss": 2.2227,
"step": 1050
},
{
"epoch": 0.3,
"learning_rate": 4.727314370749988e-05,
"loss": 2.246,
"step": 1060
},
{
"epoch": 0.3,
"learning_rate": 4.7222429865896e-05,
"loss": 2.2351,
"step": 1070
},
{
"epoch": 0.31,
"learning_rate": 4.717127649361444e-05,
"loss": 2.2326,
"step": 1080
},
{
"epoch": 0.31,
"learning_rate": 4.71196846024022e-05,
"loss": 2.2098,
"step": 1090
},
{
"epoch": 0.31,
"learning_rate": 4.7067655212679585e-05,
"loss": 2.2268,
"step": 1100
},
{
"epoch": 0.31,
"learning_rate": 4.701518935352006e-05,
"loss": 2.2108,
"step": 1110
},
{
"epoch": 0.32,
"learning_rate": 4.69622880626299e-05,
"loss": 2.2288,
"step": 1120
},
{
"epoch": 0.32,
"learning_rate": 4.690895238632762e-05,
"loss": 2.2211,
"step": 1130
},
{
"epoch": 0.32,
"learning_rate": 4.685518337952335e-05,
"loss": 2.2084,
"step": 1140
},
{
"epoch": 0.33,
"learning_rate": 4.680098210569789e-05,
"loss": 2.245,
"step": 1150
},
{
"epoch": 0.33,
"learning_rate": 4.674634963688177e-05,
"loss": 2.2657,
"step": 1160
},
{
"epoch": 0.33,
"learning_rate": 4.669128705363394e-05,
"loss": 2.2305,
"step": 1170
},
{
"epoch": 0.33,
"learning_rate": 4.663579544502049e-05,
"loss": 2.2582,
"step": 1180
},
{
"epoch": 0.34,
"learning_rate": 4.657987590859305e-05,
"loss": 2.2154,
"step": 1190
},
{
"epoch": 0.34,
"learning_rate": 4.652352955036713e-05,
"loss": 2.2047,
"step": 1200
},
{
"epoch": 0.34,
"learning_rate": 4.646675748480019e-05,
"loss": 2.2122,
"step": 1210
},
{
"epoch": 0.35,
"learning_rate": 4.640956083476964e-05,
"loss": 2.1967,
"step": 1220
},
{
"epoch": 0.35,
"learning_rate": 4.635194073155061e-05,
"loss": 2.2335,
"step": 1230
},
{
"epoch": 0.35,
"learning_rate": 4.62938983147936e-05,
"loss": 2.2196,
"step": 1240
},
{
"epoch": 0.35,
"learning_rate": 4.62354347325019e-05,
"loss": 2.2187,
"step": 1250
},
{
"epoch": 0.36,
"learning_rate": 4.6176551141008916e-05,
"loss": 2.2341,
"step": 1260
},
{
"epoch": 0.36,
"learning_rate": 4.611724870495528e-05,
"loss": 2.2043,
"step": 1270
},
{
"epoch": 0.36,
"learning_rate": 4.605752859726583e-05,
"loss": 2.2006,
"step": 1280
},
{
"epoch": 0.37,
"learning_rate": 4.5997391999126397e-05,
"loss": 2.2017,
"step": 1290
},
{
"epoch": 0.37,
"learning_rate": 4.5936840099960465e-05,
"loss": 2.2377,
"step": 1300
},
{
"epoch": 0.37,
"learning_rate": 4.587587409740561e-05,
"loss": 2.1889,
"step": 1310
},
{
"epoch": 0.37,
"learning_rate": 4.581449519728986e-05,
"loss": 2.202,
"step": 1320
},
{
"epoch": 0.38,
"learning_rate": 4.575270461360779e-05,
"loss": 2.2038,
"step": 1330
},
{
"epoch": 0.38,
"learning_rate": 4.569050356849655e-05,
"loss": 2.1918,
"step": 1340
},
{
"epoch": 0.38,
"learning_rate": 4.562789329221169e-05,
"loss": 2.2356,
"step": 1350
},
{
"epoch": 0.39,
"learning_rate": 4.5564875023102796e-05,
"loss": 2.2356,
"step": 1360
},
{
"epoch": 0.39,
"learning_rate": 4.550145000758905e-05,
"loss": 2.2383,
"step": 1370
},
{
"epoch": 0.39,
"learning_rate": 4.543761950013453e-05,
"loss": 2.238,
"step": 1380
},
{
"epoch": 0.39,
"learning_rate": 4.537338476322341e-05,
"loss": 2.2153,
"step": 1390
},
{
"epoch": 0.4,
"learning_rate": 4.530874706733502e-05,
"loss": 2.2484,
"step": 1400
},
{
"epoch": 0.4,
"learning_rate": 4.524370769091866e-05,
"loss": 2.204,
"step": 1410
},
{
"epoch": 0.4,
"learning_rate": 4.5178267920368366e-05,
"loss": 2.2408,
"step": 1420
},
{
"epoch": 0.4,
"learning_rate": 4.511242904999746e-05,
"loss": 2.2172,
"step": 1430
},
{
"epoch": 0.41,
"learning_rate": 4.50461923820129e-05,
"loss": 2.2467,
"step": 1440
},
{
"epoch": 0.41,
"learning_rate": 4.497955922648961e-05,
"loss": 2.2417,
"step": 1450
},
{
"epoch": 0.41,
"learning_rate": 4.491253090134447e-05,
"loss": 2.1996,
"step": 1460
},
{
"epoch": 0.42,
"learning_rate": 4.4845108732310355e-05,
"loss": 2.2065,
"step": 1470
},
{
"epoch": 0.42,
"learning_rate": 4.4777294052909804e-05,
"loss": 2.2379,
"step": 1480
},
{
"epoch": 0.42,
"learning_rate": 4.470908820442873e-05,
"loss": 2.2136,
"step": 1490
},
{
"epoch": 0.42,
"learning_rate": 4.464049253588985e-05,
"loss": 2.1988,
"step": 1500
},
{
"epoch": 0.43,
"learning_rate": 4.457150840402604e-05,
"loss": 2.2133,
"step": 1510
},
{
"epoch": 0.43,
"learning_rate": 4.450213717325343e-05,
"loss": 2.2173,
"step": 1520
},
{
"epoch": 0.43,
"learning_rate": 4.443238021564448e-05,
"loss": 2.2417,
"step": 1530
},
{
"epoch": 0.44,
"learning_rate": 4.4362238910900835e-05,
"loss": 2.2359,
"step": 1540
},
{
"epoch": 0.44,
"learning_rate": 4.4291714646326024e-05,
"loss": 2.1843,
"step": 1550
},
{
"epoch": 0.44,
"learning_rate": 4.4220808816797995e-05,
"loss": 2.2367,
"step": 1560
},
{
"epoch": 0.44,
"learning_rate": 4.414952282474158e-05,
"loss": 2.2484,
"step": 1570
},
{
"epoch": 0.45,
"learning_rate": 4.407785808010073e-05,
"loss": 2.2269,
"step": 1580
},
{
"epoch": 0.45,
"learning_rate": 4.400581600031062e-05,
"loss": 2.2405,
"step": 1590
},
{
"epoch": 0.45,
"learning_rate": 4.3933398010269614e-05,
"loss": 2.1853,
"step": 1600
},
{
"epoch": 0.46,
"learning_rate": 4.386060554231111e-05,
"loss": 2.1855,
"step": 1610
},
{
"epoch": 0.46,
"learning_rate": 4.378744003617517e-05,
"loss": 2.1776,
"step": 1620
},
{
"epoch": 0.46,
"learning_rate": 4.371390293898007e-05,
"loss": 2.198,
"step": 1630
},
{
"epoch": 0.46,
"learning_rate": 4.3639995705193684e-05,
"loss": 2.2166,
"step": 1640
},
{
"epoch": 0.47,
"learning_rate": 4.3565719796604706e-05,
"loss": 2.1939,
"step": 1650
},
{
"epoch": 0.47,
"learning_rate": 4.3491076682293715e-05,
"loss": 2.1843,
"step": 1660
},
{
"epoch": 0.47,
"learning_rate": 4.341606783860416e-05,
"loss": 2.1955,
"step": 1670
},
{
"epoch": 0.48,
"learning_rate": 4.334069474911313e-05,
"loss": 2.1792,
"step": 1680
},
{
"epoch": 0.48,
"learning_rate": 4.326495890460204e-05,
"loss": 2.1898,
"step": 1690
},
{
"epoch": 0.48,
"learning_rate": 4.3188861803027094e-05,
"loss": 2.1906,
"step": 1700
},
{
"epoch": 0.48,
"learning_rate": 4.3112404949489715e-05,
"loss": 2.2328,
"step": 1710
},
{
"epoch": 0.49,
"learning_rate": 4.303558985620673e-05,
"loss": 2.2169,
"step": 1720
},
{
"epoch": 0.49,
"learning_rate": 4.2958418042480506e-05,
"loss": 2.1611,
"step": 1730
},
{
"epoch": 0.49,
"learning_rate": 4.288089103466886e-05,
"loss": 2.2076,
"step": 1740
},
{
"epoch": 0.5,
"learning_rate": 4.280301036615489e-05,
"loss": 2.1789,
"step": 1750
},
{
"epoch": 0.5,
"learning_rate": 4.272477757731662e-05,
"loss": 2.2005,
"step": 1760
},
{
"epoch": 0.5,
"learning_rate": 4.2646194215496595e-05,
"loss": 2.1777,
"step": 1770
},
{
"epoch": 0.5,
"learning_rate": 4.2567261834971206e-05,
"loss": 2.1725,
"step": 1780
},
{
"epoch": 0.51,
"learning_rate": 4.248798199691998e-05,
"loss": 2.2078,
"step": 1790
},
{
"epoch": 0.51,
"learning_rate": 4.2408356269394714e-05,
"loss": 2.2008,
"step": 1800
},
{
"epoch": 0.51,
"learning_rate": 4.2328386227288445e-05,
"loss": 2.1663,
"step": 1810
},
{
"epoch": 0.52,
"learning_rate": 4.224807345230429e-05,
"loss": 2.2049,
"step": 1820
},
{
"epoch": 0.52,
"learning_rate": 4.21674195329242e-05,
"loss": 2.1745,
"step": 1830
},
{
"epoch": 0.52,
"learning_rate": 4.208642606437748e-05,
"loss": 2.2407,
"step": 1840
},
{
"epoch": 0.52,
"learning_rate": 4.2005094648609325e-05,
"loss": 2.2048,
"step": 1850
},
{
"epoch": 0.53,
"learning_rate": 4.192342689424903e-05,
"loss": 2.1035,
"step": 1860
},
{
"epoch": 0.53,
"learning_rate": 4.184142441657827e-05,
"loss": 2.2034,
"step": 1870
},
{
"epoch": 0.53,
"learning_rate": 4.1759088837499095e-05,
"loss": 2.2424,
"step": 1880
},
{
"epoch": 0.54,
"learning_rate": 4.167642178550186e-05,
"loss": 2.2036,
"step": 1890
},
{
"epoch": 0.54,
"learning_rate": 4.159342489563304e-05,
"loss": 2.1925,
"step": 1900
},
{
"epoch": 0.54,
"learning_rate": 4.151009980946287e-05,
"loss": 2.1977,
"step": 1910
},
{
"epoch": 0.54,
"learning_rate": 4.142644817505284e-05,
"loss": 2.1568,
"step": 1920
},
{
"epoch": 0.55,
"learning_rate": 4.1342471646923196e-05,
"loss": 2.2062,
"step": 1930
},
{
"epoch": 0.55,
"learning_rate": 4.1258171886020124e-05,
"loss": 2.1841,
"step": 1940
},
{
"epoch": 0.55,
"learning_rate": 4.117355055968295e-05,
"loss": 2.181,
"step": 1950
},
{
"epoch": 0.55,
"learning_rate": 4.1088609341611135e-05,
"loss": 2.1584,
"step": 1960
},
{
"epoch": 0.56,
"learning_rate": 4.1003349911831166e-05,
"loss": 2.2113,
"step": 1970
},
{
"epoch": 0.56,
"learning_rate": 4.091777395666338e-05,
"loss": 2.235,
"step": 1980
},
{
"epoch": 0.56,
"learning_rate": 4.0831883168688534e-05,
"loss": 2.1844,
"step": 1990
},
{
"epoch": 0.57,
"learning_rate": 4.07456792467144e-05,
"loss": 2.2042,
"step": 2000
},
{
"epoch": 0.57,
"learning_rate": 4.065916389574211e-05,
"loss": 2.2178,
"step": 2010
},
{
"epoch": 0.57,
"learning_rate": 4.057233882693246e-05,
"loss": 2.1875,
"step": 2020
},
{
"epoch": 0.57,
"learning_rate": 4.048520575757207e-05,
"loss": 2.1612,
"step": 2030
},
{
"epoch": 0.58,
"learning_rate": 4.0397766411039395e-05,
"loss": 2.1706,
"step": 2040
},
{
"epoch": 0.58,
"learning_rate": 4.031002251677067e-05,
"loss": 2.1786,
"step": 2050
},
{
"epoch": 0.58,
"learning_rate": 4.022197581022567e-05,
"loss": 2.1754,
"step": 2060
},
{
"epoch": 0.59,
"learning_rate": 4.0133628032853435e-05,
"loss": 2.1893,
"step": 2070
},
{
"epoch": 0.59,
"learning_rate": 4.004498093205773e-05,
"loss": 2.142,
"step": 2080
},
{
"epoch": 0.59,
"learning_rate": 3.99560362611626e-05,
"loss": 2.2265,
"step": 2090
},
{
"epoch": 0.59,
"learning_rate": 3.986679577937762e-05,
"loss": 2.1621,
"step": 2100
},
{
"epoch": 0.6,
"learning_rate": 3.9777261251763145e-05,
"loss": 2.1935,
"step": 2110
},
{
"epoch": 0.6,
"learning_rate": 3.9687434449195316e-05,
"loss": 2.2049,
"step": 2120
},
{
"epoch": 0.6,
"learning_rate": 3.959731714833115e-05,
"loss": 2.179,
"step": 2130
},
{
"epoch": 0.61,
"learning_rate": 3.95069111315733e-05,
"loss": 2.1834,
"step": 2140
},
{
"epoch": 0.61,
"learning_rate": 3.9416218187034876e-05,
"loss": 2.2075,
"step": 2150
},
{
"epoch": 0.61,
"learning_rate": 3.9325240108504016e-05,
"loss": 2.2044,
"step": 2160
},
{
"epoch": 0.61,
"learning_rate": 3.923397869540845e-05,
"loss": 2.1615,
"step": 2170
},
{
"epoch": 0.62,
"learning_rate": 3.91424357527799e-05,
"loss": 2.1678,
"step": 2180
},
{
"epoch": 0.62,
"learning_rate": 3.905061309121837e-05,
"loss": 2.1597,
"step": 2190
},
{
"epoch": 0.62,
"learning_rate": 3.895851252685634e-05,
"loss": 2.1487,
"step": 2200
},
{
"epoch": 0.63,
"learning_rate": 3.886613588132284e-05,
"loss": 2.2268,
"step": 2210
},
{
"epoch": 0.63,
"learning_rate": 3.877348498170743e-05,
"loss": 2.1913,
"step": 2220
},
{
"epoch": 0.63,
"learning_rate": 3.868056166052407e-05,
"loss": 2.185,
"step": 2230
},
{
"epoch": 0.63,
"learning_rate": 3.858736775567484e-05,
"loss": 2.1779,
"step": 2240
},
{
"epoch": 0.64,
"learning_rate": 3.8493905110413626e-05,
"loss": 2.1733,
"step": 2250
},
{
"epoch": 0.64,
"learning_rate": 3.8400175573309646e-05,
"loss": 2.1982,
"step": 2260
},
{
"epoch": 0.64,
"learning_rate": 3.83061809982109e-05,
"loss": 2.1871,
"step": 2270
},
{
"epoch": 0.65,
"learning_rate": 3.821192324420747e-05,
"loss": 2.2081,
"step": 2280
},
{
"epoch": 0.65,
"learning_rate": 3.8117404175594795e-05,
"loss": 2.2318,
"step": 2290
},
{
"epoch": 0.65,
"learning_rate": 3.8022625661836775e-05,
"loss": 2.1611,
"step": 2300
},
{
"epoch": 0.65,
"learning_rate": 3.792758957752879e-05,
"loss": 2.1715,
"step": 2310
},
{
"epoch": 0.66,
"learning_rate": 3.783229780236062e-05,
"loss": 2.1816,
"step": 2320
},
{
"epoch": 0.66,
"learning_rate": 3.773675222107929e-05,
"loss": 2.1705,
"step": 2330
},
{
"epoch": 0.66,
"learning_rate": 3.7640954723451784e-05,
"loss": 2.1688,
"step": 2340
},
{
"epoch": 0.67,
"learning_rate": 3.7544907204227646e-05,
"loss": 2.1511,
"step": 2350
},
{
"epoch": 0.67,
"learning_rate": 3.7448611563101536e-05,
"loss": 2.1671,
"step": 2360
},
{
"epoch": 0.67,
"learning_rate": 3.735206970467564e-05,
"loss": 2.1663,
"step": 2370
},
{
"epoch": 0.67,
"learning_rate": 3.7255283538421984e-05,
"loss": 2.1824,
"step": 2380
},
{
"epoch": 0.68,
"learning_rate": 3.7158254978644703e-05,
"loss": 2.1788,
"step": 2390
},
{
"epoch": 0.68,
"learning_rate": 3.706098594444216e-05,
"loss": 2.1394,
"step": 2400
},
{
"epoch": 0.68,
"learning_rate": 3.6963478359668966e-05,
"loss": 2.2013,
"step": 2410
},
{
"epoch": 0.69,
"learning_rate": 3.6865734152897987e-05,
"loss": 2.1638,
"step": 2420
},
{
"epoch": 0.69,
"learning_rate": 3.676775525738213e-05,
"loss": 2.1814,
"step": 2430
},
{
"epoch": 0.69,
"learning_rate": 3.666954361101616e-05,
"loss": 2.1763,
"step": 2440
},
{
"epoch": 0.69,
"learning_rate": 3.657110115629832e-05,
"loss": 2.1531,
"step": 2450
},
{
"epoch": 0.7,
"learning_rate": 3.647242984029201e-05,
"loss": 2.1862,
"step": 2460
},
{
"epoch": 0.7,
"learning_rate": 3.637353161458713e-05,
"loss": 2.1611,
"step": 2470
},
{
"epoch": 0.7,
"learning_rate": 3.627440843526158e-05,
"loss": 2.1733,
"step": 2480
},
{
"epoch": 0.7,
"learning_rate": 3.6175062262842585e-05,
"loss": 2.1816,
"step": 2490
},
{
"epoch": 0.71,
"learning_rate": 3.607549506226784e-05,
"loss": 2.1653,
"step": 2500
},
{
"epoch": 0.71,
"learning_rate": 3.5975708802846725e-05,
"loss": 2.1777,
"step": 2510
},
{
"epoch": 0.71,
"learning_rate": 3.587570545822128e-05,
"loss": 2.189,
"step": 2520
},
{
"epoch": 0.72,
"learning_rate": 3.5775487006327215e-05,
"loss": 2.1875,
"step": 2530
},
{
"epoch": 0.72,
"learning_rate": 3.567505542935482e-05,
"loss": 2.1751,
"step": 2540
},
{
"epoch": 0.72,
"learning_rate": 3.557441271370968e-05,
"loss": 2.1778,
"step": 2550
},
{
"epoch": 0.72,
"learning_rate": 3.547356084997344e-05,
"loss": 2.1934,
"step": 2560
},
{
"epoch": 0.73,
"learning_rate": 3.537250183286443e-05,
"loss": 2.1308,
"step": 2570
},
{
"epoch": 0.73,
"learning_rate": 3.527123766119818e-05,
"loss": 2.1722,
"step": 2580
},
{
"epoch": 0.73,
"learning_rate": 3.5169770337847935e-05,
"loss": 2.1744,
"step": 2590
},
{
"epoch": 0.74,
"learning_rate": 3.5068101869705004e-05,
"loss": 2.2074,
"step": 2600
},
{
"epoch": 0.74,
"learning_rate": 3.496623426763908e-05,
"loss": 2.1795,
"step": 2610
},
{
"epoch": 0.74,
"learning_rate": 3.486416954645846e-05,
"loss": 2.1539,
"step": 2620
},
{
"epoch": 0.74,
"learning_rate": 3.4761909724870205e-05,
"loss": 2.1723,
"step": 2630
},
{
"epoch": 0.75,
"learning_rate": 3.4659456825440216e-05,
"loss": 2.1858,
"step": 2640
},
{
"epoch": 0.75,
"learning_rate": 3.4556812874553204e-05,
"loss": 2.199,
"step": 2650
},
{
"epoch": 0.75,
"learning_rate": 3.445397990237264e-05,
"loss": 2.202,
"step": 2660
},
{
"epoch": 0.76,
"learning_rate": 3.4350959942800584e-05,
"loss": 2.1559,
"step": 2670
},
{
"epoch": 0.76,
"learning_rate": 3.424775503343747e-05,
"loss": 2.1746,
"step": 2680
},
{
"epoch": 0.76,
"learning_rate": 3.414436721554179e-05,
"loss": 2.1608,
"step": 2690
},
{
"epoch": 0.76,
"learning_rate": 3.4040798533989745e-05,
"loss": 2.1568,
"step": 2700
},
{
"epoch": 0.77,
"learning_rate": 3.3937051037234745e-05,
"loss": 2.1469,
"step": 2710
},
{
"epoch": 0.77,
"learning_rate": 3.3833126777266974e-05,
"loss": 2.1815,
"step": 2720
},
{
"epoch": 0.77,
"learning_rate": 3.3729027809572735e-05,
"loss": 2.1938,
"step": 2730
},
{
"epoch": 0.78,
"learning_rate": 3.3624756193093835e-05,
"loss": 2.1436,
"step": 2740
},
{
"epoch": 0.78,
"learning_rate": 3.3520313990186856e-05,
"loss": 2.1679,
"step": 2750
},
{
"epoch": 0.78,
"learning_rate": 3.341570326658232e-05,
"loss": 2.1419,
"step": 2760
},
{
"epoch": 0.78,
"learning_rate": 3.331092609134392e-05,
"loss": 2.1654,
"step": 2770
},
{
"epoch": 0.79,
"learning_rate": 3.32059845368275e-05,
"loss": 2.1549,
"step": 2780
},
{
"epoch": 0.79,
"learning_rate": 3.310088067864015e-05,
"loss": 2.1673,
"step": 2790
},
{
"epoch": 0.79,
"learning_rate": 3.299561659559909e-05,
"loss": 2.1571,
"step": 2800
},
{
"epoch": 0.8,
"learning_rate": 3.289019436969057e-05,
"loss": 2.1475,
"step": 2810
},
{
"epoch": 0.8,
"learning_rate": 3.278461608602873e-05,
"loss": 2.171,
"step": 2820
},
{
"epoch": 0.8,
"learning_rate": 3.267888383281431e-05,
"loss": 2.1774,
"step": 2830
},
{
"epoch": 0.8,
"learning_rate": 3.257299970129336e-05,
"loss": 2.1624,
"step": 2840
},
{
"epoch": 0.81,
"learning_rate": 3.246696578571592e-05,
"loss": 2.1492,
"step": 2850
},
{
"epoch": 0.81,
"learning_rate": 3.236078418329452e-05,
"loss": 2.2018,
"step": 2860
},
{
"epoch": 0.81,
"learning_rate": 3.225445699416276e-05,
"loss": 2.1521,
"step": 2870
},
{
"epoch": 0.82,
"learning_rate": 3.214798632133376e-05,
"loss": 2.1719,
"step": 2880
},
{
"epoch": 0.82,
"learning_rate": 3.204137427065857e-05,
"loss": 2.1807,
"step": 2890
},
{
"epoch": 0.82,
"learning_rate": 3.193462295078448e-05,
"loss": 2.1755,
"step": 2900
},
{
"epoch": 0.82,
"learning_rate": 3.1827734473113356e-05,
"loss": 2.1715,
"step": 2910
},
{
"epoch": 0.83,
"learning_rate": 3.172071095175989e-05,
"loss": 2.1799,
"step": 2920
},
{
"epoch": 0.83,
"learning_rate": 3.1613554503509715e-05,
"loss": 2.1348,
"step": 2930
},
{
"epoch": 0.83,
"learning_rate": 3.1506267247777645e-05,
"loss": 2.1886,
"step": 2940
},
{
"epoch": 0.84,
"learning_rate": 3.139885130656563e-05,
"loss": 2.1249,
"step": 2950
},
{
"epoch": 0.84,
"learning_rate": 3.129130880442091e-05,
"loss": 2.1562,
"step": 2960
},
{
"epoch": 0.84,
"learning_rate": 3.118364186839389e-05,
"loss": 2.1613,
"step": 2970
},
{
"epoch": 0.84,
"learning_rate": 3.107585262799613e-05,
"loss": 2.1353,
"step": 2980
},
{
"epoch": 0.85,
"learning_rate": 3.096794321515823e-05,
"loss": 2.2067,
"step": 2990
},
{
"epoch": 0.85,
"learning_rate": 3.0859915764187616e-05,
"loss": 2.1776,
"step": 3000
},
{
"epoch": 0.85,
"learning_rate": 3.075177241172637e-05,
"loss": 2.1759,
"step": 3010
},
{
"epoch": 0.86,
"learning_rate": 3.064351529670895e-05,
"loss": 2.1333,
"step": 3020
},
{
"epoch": 0.86,
"learning_rate": 3.0535146560319895e-05,
"loss": 2.1873,
"step": 3030
},
{
"epoch": 0.86,
"learning_rate": 3.0426668345951448e-05,
"loss": 2.197,
"step": 3040
},
{
"epoch": 0.86,
"learning_rate": 3.031808279916121e-05,
"loss": 2.2196,
"step": 3050
},
{
"epoch": 0.87,
"learning_rate": 3.0209392067629676e-05,
"loss": 2.1505,
"step": 3060
},
{
"epoch": 0.87,
"learning_rate": 3.010059830111773e-05,
"loss": 2.1615,
"step": 3070
},
{
"epoch": 0.87,
"learning_rate": 2.9991703651424184e-05,
"loss": 2.1979,
"step": 3080
},
{
"epoch": 0.87,
"learning_rate": 2.988271027234317e-05,
"loss": 2.1664,
"step": 3090
},
{
"epoch": 0.88,
"learning_rate": 2.977362031962158e-05,
"loss": 2.168,
"step": 3100
},
{
"epoch": 0.88,
"learning_rate": 2.966443595091637e-05,
"loss": 2.19,
"step": 3110
},
{
"epoch": 0.88,
"learning_rate": 2.9555159325751947e-05,
"loss": 2.1612,
"step": 3120
},
{
"epoch": 0.89,
"learning_rate": 2.9445792605477433e-05,
"loss": 2.1514,
"step": 3130
},
{
"epoch": 0.89,
"learning_rate": 2.933633795322391e-05,
"loss": 2.1799,
"step": 3140
},
{
"epoch": 0.89,
"learning_rate": 2.9226797533861615e-05,
"loss": 2.1733,
"step": 3150
},
{
"epoch": 0.89,
"learning_rate": 2.9117173513957186e-05,
"loss": 2.1627,
"step": 3160
},
{
"epoch": 0.9,
"learning_rate": 2.900746806173074e-05,
"loss": 2.1348,
"step": 3170
},
{
"epoch": 0.9,
"learning_rate": 2.889768334701305e-05,
"loss": 2.1721,
"step": 3180
},
{
"epoch": 0.9,
"learning_rate": 2.8787821541202564e-05,
"loss": 2.1607,
"step": 3190
},
{
"epoch": 0.91,
"learning_rate": 2.8677884817222505e-05,
"loss": 2.1716,
"step": 3200
},
{
"epoch": 0.91,
"learning_rate": 2.856787534947789e-05,
"loss": 2.1683,
"step": 3210
},
{
"epoch": 0.91,
"learning_rate": 2.8457795313812507e-05,
"loss": 2.1815,
"step": 3220
},
{
"epoch": 0.91,
"learning_rate": 2.8347646887465867e-05,
"loss": 2.136,
"step": 3230
},
{
"epoch": 0.92,
"learning_rate": 2.8237432249030183e-05,
"loss": 2.1496,
"step": 3240
},
{
"epoch": 0.92,
"learning_rate": 2.812715357840725e-05,
"loss": 2.1506,
"step": 3250
},
{
"epoch": 0.92,
"learning_rate": 2.801681305676534e-05,
"loss": 2.1815,
"step": 3260
},
{
"epoch": 0.93,
"learning_rate": 2.7906412866496034e-05,
"loss": 2.149,
"step": 3270
},
{
"epoch": 0.93,
"learning_rate": 2.7795955191171114e-05,
"loss": 2.1821,
"step": 3280
},
{
"epoch": 0.93,
"learning_rate": 2.7685442215499323e-05,
"loss": 2.1704,
"step": 3290
},
{
"epoch": 0.93,
"learning_rate": 2.7574876125283176e-05,
"loss": 2.1578,
"step": 3300
},
{
"epoch": 0.94,
"learning_rate": 2.7464259107375722e-05,
"loss": 2.1582,
"step": 3310
},
{
"epoch": 0.94,
"learning_rate": 2.7353593349637297e-05,
"loss": 2.1623,
"step": 3320
},
{
"epoch": 0.94,
"learning_rate": 2.7242881040892255e-05,
"loss": 2.181,
"step": 3330
},
{
"epoch": 0.95,
"learning_rate": 2.713212437088564e-05,
"loss": 2.1829,
"step": 3340
},
{
"epoch": 0.95,
"learning_rate": 2.702132553023993e-05,
"loss": 2.1708,
"step": 3350
},
{
"epoch": 0.95,
"learning_rate": 2.691048671041169e-05,
"loss": 2.1588,
"step": 3360
},
{
"epoch": 0.95,
"learning_rate": 2.6799610103648197e-05,
"loss": 2.1266,
"step": 3370
},
{
"epoch": 0.96,
"learning_rate": 2.668869790294411e-05,
"loss": 2.1572,
"step": 3380
},
{
"epoch": 0.96,
"learning_rate": 2.6577752301998098e-05,
"loss": 2.1758,
"step": 3390
},
{
"epoch": 0.96,
"learning_rate": 2.6466775495169437e-05,
"loss": 2.177,
"step": 3400
},
{
"epoch": 0.97,
"learning_rate": 2.6355769677434628e-05,
"loss": 2.1471,
"step": 3410
},
{
"epoch": 0.97,
"learning_rate": 2.6244737044343954e-05,
"loss": 2.1505,
"step": 3420
},
{
"epoch": 0.97,
"learning_rate": 2.6133679791978065e-05,
"loss": 2.1668,
"step": 3430
},
{
"epoch": 0.97,
"learning_rate": 2.602260011690457e-05,
"loss": 2.171,
"step": 3440
},
{
"epoch": 0.98,
"learning_rate": 2.591150021613456e-05,
"loss": 2.122,
"step": 3450
},
{
"epoch": 0.98,
"learning_rate": 2.580038228707916e-05,
"loss": 2.1541,
"step": 3460
},
{
"epoch": 0.98,
"learning_rate": 2.568924852750607e-05,
"loss": 2.1942,
"step": 3470
},
{
"epoch": 0.99,
"learning_rate": 2.55781011354961e-05,
"loss": 2.1709,
"step": 3480
},
{
"epoch": 0.99,
"learning_rate": 2.5466942309399706e-05,
"loss": 2.1376,
"step": 3490
},
{
"epoch": 0.99,
"learning_rate": 2.535577424779347e-05,
"loss": 2.1503,
"step": 3500
},
{
"epoch": 0.99,
"learning_rate": 2.5244599149436642e-05,
"loss": 2.1259,
"step": 3510
},
{
"epoch": 1.0,
"learning_rate": 2.5133419213227684e-05,
"loss": 2.1691,
"step": 3520
},
{
"epoch": 1.0,
"learning_rate": 2.5022236638160712e-05,
"loss": 2.1703,
"step": 3530
},
{
"epoch": 1.0,
"learning_rate": 2.491105362328204e-05,
"loss": 2.1584,
"step": 3540
},
{
"epoch": 1.01,
"learning_rate": 2.4799872367646685e-05,
"loss": 2.2039,
"step": 3550
},
{
"epoch": 1.01,
"learning_rate": 2.468869507027486e-05,
"loss": 2.1016,
"step": 3560
},
{
"epoch": 1.01,
"learning_rate": 2.457752393010851e-05,
"loss": 2.1971,
"step": 3570
},
{
"epoch": 1.01,
"learning_rate": 2.4466361145967782e-05,
"loss": 2.1707,
"step": 3580
},
{
"epoch": 1.02,
"learning_rate": 2.4355208916507554e-05,
"loss": 2.0819,
"step": 3590
},
{
"epoch": 1.02,
"learning_rate": 2.4244069440173938e-05,
"loss": 2.1181,
"step": 3600
},
{
"epoch": 1.02,
"learning_rate": 2.413294491516083e-05,
"loss": 2.1788,
"step": 3610
},
{
"epoch": 1.02,
"learning_rate": 2.402183753936639e-05,
"loss": 2.1672,
"step": 3620
},
{
"epoch": 1.03,
"learning_rate": 2.3910749510349574e-05,
"loss": 2.1527,
"step": 3630
},
{
"epoch": 1.03,
"learning_rate": 2.3799683025286727e-05,
"loss": 2.1734,
"step": 3640
},
{
"epoch": 1.03,
"learning_rate": 2.3688640280928043e-05,
"loss": 2.1539,
"step": 3650
},
{
"epoch": 1.04,
"learning_rate": 2.357762347355418e-05,
"loss": 2.1398,
"step": 3660
},
{
"epoch": 1.04,
"learning_rate": 2.346663479893277e-05,
"loss": 2.1568,
"step": 3670
},
{
"epoch": 1.04,
"learning_rate": 2.3355676452275057e-05,
"loss": 2.1839,
"step": 3680
},
{
"epoch": 1.04,
"learning_rate": 2.3244750628192395e-05,
"loss": 2.13,
"step": 3690
},
{
"epoch": 1.05,
"learning_rate": 2.313385952065291e-05,
"loss": 2.106,
"step": 3700
},
{
"epoch": 1.05,
"learning_rate": 2.302300532293806e-05,
"loss": 2.1677,
"step": 3710
},
{
"epoch": 1.05,
"learning_rate": 2.2912190227599305e-05,
"loss": 2.1488,
"step": 3720
},
{
"epoch": 1.06,
"learning_rate": 2.280141642641468e-05,
"loss": 2.0953,
"step": 3730
},
{
"epoch": 1.06,
"learning_rate": 2.26906861103455e-05,
"loss": 2.1371,
"step": 3740
},
{
"epoch": 1.06,
"learning_rate": 2.2580001469492968e-05,
"loss": 2.1571,
"step": 3750
},
{
"epoch": 1.06,
"learning_rate": 2.2469364693054938e-05,
"loss": 2.1491,
"step": 3760
},
{
"epoch": 1.07,
"learning_rate": 2.235877796928252e-05,
"loss": 2.1469,
"step": 3770
},
{
"epoch": 1.07,
"learning_rate": 2.2248243485436892e-05,
"loss": 2.1659,
"step": 3780
},
{
"epoch": 1.07,
"learning_rate": 2.213776342774595e-05,
"loss": 2.1426,
"step": 3790
},
{
"epoch": 1.08,
"learning_rate": 2.2027339981361156e-05,
"loss": 2.1565,
"step": 3800
},
{
"epoch": 1.08,
"learning_rate": 2.1916975330314235e-05,
"loss": 2.1913,
"step": 3810
},
{
"epoch": 1.08,
"learning_rate": 2.1806671657474048e-05,
"loss": 2.0965,
"step": 3820
},
{
"epoch": 1.08,
"learning_rate": 2.1696431144503356e-05,
"loss": 2.1516,
"step": 3830
},
{
"epoch": 1.09,
"learning_rate": 2.1586255971815738e-05,
"loss": 2.1558,
"step": 3840
},
{
"epoch": 1.09,
"learning_rate": 2.147614831853239e-05,
"loss": 2.1093,
"step": 3850
},
{
"epoch": 1.09,
"learning_rate": 2.136611036243908e-05,
"loss": 2.1752,
"step": 3860
},
{
"epoch": 1.1,
"learning_rate": 2.1256144279943066e-05,
"loss": 2.1185,
"step": 3870
},
{
"epoch": 1.1,
"learning_rate": 2.114625224603002e-05,
"loss": 2.1106,
"step": 3880
},
{
"epoch": 1.1,
"learning_rate": 2.1036436434221048e-05,
"loss": 2.1949,
"step": 3890
},
{
"epoch": 1.1,
"learning_rate": 2.0926699016529656e-05,
"loss": 2.1041,
"step": 3900
},
{
"epoch": 1.11,
"learning_rate": 2.0817042163418858e-05,
"loss": 2.1484,
"step": 3910
},
{
"epoch": 1.11,
"learning_rate": 2.0707468043758156e-05,
"loss": 2.1623,
"step": 3920
},
{
"epoch": 1.11,
"learning_rate": 2.059797882478074e-05,
"loss": 2.1681,
"step": 3930
},
{
"epoch": 1.12,
"learning_rate": 2.048857667204053e-05,
"loss": 2.1909,
"step": 3940
},
{
"epoch": 1.12,
"learning_rate": 2.0379263749369434e-05,
"loss": 2.1334,
"step": 3950
},
{
"epoch": 1.12,
"learning_rate": 2.027004221883447e-05,
"loss": 2.103,
"step": 3960
},
{
"epoch": 1.12,
"learning_rate": 2.016091424069506e-05,
"loss": 2.1151,
"step": 3970
},
{
"epoch": 1.13,
"learning_rate": 2.0051881973360266e-05,
"loss": 2.1309,
"step": 3980
},
{
"epoch": 1.13,
"learning_rate": 1.9942947573346133e-05,
"loss": 2.1387,
"step": 3990
},
{
"epoch": 1.13,
"learning_rate": 1.9834113195233e-05,
"loss": 2.1635,
"step": 4000
},
{
"epoch": 1.14,
"learning_rate": 1.9725380991622923e-05,
"loss": 2.1245,
"step": 4010
},
{
"epoch": 1.14,
"learning_rate": 1.961675311309705e-05,
"loss": 2.1451,
"step": 4020
},
{
"epoch": 1.14,
"learning_rate": 1.951907899616401e-05,
"loss": 2.0919,
"step": 4030
},
{
"epoch": 1.14,
"learning_rate": 1.9410655252724894e-05,
"loss": 2.1091,
"step": 4040
},
{
"epoch": 1.15,
"learning_rate": 1.9302342059234615e-05,
"loss": 2.1417,
"step": 4050
},
{
"epoch": 1.15,
"learning_rate": 1.9194141557986852e-05,
"loss": 2.1835,
"step": 4060
},
{
"epoch": 1.15,
"learning_rate": 1.9086055889046402e-05,
"loss": 2.1797,
"step": 4070
},
{
"epoch": 1.16,
"learning_rate": 1.8978087190206803e-05,
"loss": 2.1279,
"step": 4080
},
{
"epoch": 1.16,
"learning_rate": 1.8870237596948115e-05,
"loss": 2.1245,
"step": 4090
},
{
"epoch": 1.16,
"learning_rate": 1.8762509242394602e-05,
"loss": 2.1262,
"step": 4100
},
{
"epoch": 1.16,
"learning_rate": 1.865490425727262e-05,
"loss": 2.1501,
"step": 4110
},
{
"epoch": 1.17,
"learning_rate": 1.854742476986841e-05,
"loss": 2.1359,
"step": 4120
},
{
"epoch": 1.17,
"learning_rate": 1.844007290598606e-05,
"loss": 2.1588,
"step": 4130
},
{
"epoch": 1.17,
"learning_rate": 1.8332850788905408e-05,
"loss": 2.152,
"step": 4140
},
{
"epoch": 1.17,
"learning_rate": 1.8225760539340086e-05,
"loss": 2.1351,
"step": 4150
},
{
"epoch": 1.18,
"learning_rate": 1.8118804275395535e-05,
"loss": 2.1458,
"step": 4160
},
{
"epoch": 1.18,
"learning_rate": 1.801198411252717e-05,
"loss": 2.1081,
"step": 4170
},
{
"epoch": 1.18,
"learning_rate": 1.7905302163498472e-05,
"loss": 2.1935,
"step": 4180
},
{
"epoch": 1.19,
"learning_rate": 1.7798760538339254e-05,
"loss": 2.1386,
"step": 4190
},
{
"epoch": 1.19,
"learning_rate": 1.7692361344303887e-05,
"loss": 2.1212,
"step": 4200
},
{
"epoch": 1.19,
"learning_rate": 1.7586106685829666e-05,
"loss": 2.1329,
"step": 4210
},
{
"epoch": 1.19,
"learning_rate": 1.7479998664495135e-05,
"loss": 2.1593,
"step": 4220
},
{
"epoch": 1.2,
"learning_rate": 1.7374039378978563e-05,
"loss": 2.127,
"step": 4230
},
{
"epoch": 1.2,
"learning_rate": 1.7268230925016405e-05,
"loss": 2.1485,
"step": 4240
},
{
"epoch": 1.2,
"learning_rate": 1.7162575395361877e-05,
"loss": 2.1437,
"step": 4250
},
{
"epoch": 1.21,
"learning_rate": 1.7057074879743556e-05,
"loss": 2.1413,
"step": 4260
},
{
"epoch": 1.21,
"learning_rate": 1.6951731464824018e-05,
"loss": 2.1343,
"step": 4270
},
{
"epoch": 1.21,
"learning_rate": 1.684654723415863e-05,
"loss": 2.128,
"step": 4280
},
{
"epoch": 1.21,
"learning_rate": 1.6741524268154264e-05,
"loss": 2.1334,
"step": 4290
},
{
"epoch": 1.22,
"learning_rate": 1.6636664644028214e-05,
"loss": 2.1437,
"step": 4300
},
{
"epoch": 1.22,
"learning_rate": 1.653197043576706e-05,
"loss": 2.1344,
"step": 4310
},
{
"epoch": 1.22,
"learning_rate": 1.642744371408571e-05,
"loss": 2.0975,
"step": 4320
},
{
"epoch": 1.23,
"learning_rate": 1.6323086546386363e-05,
"loss": 2.1557,
"step": 4330
},
{
"epoch": 1.23,
"learning_rate": 1.6218900996717697e-05,
"loss": 2.1278,
"step": 4340
},
{
"epoch": 1.23,
"learning_rate": 1.6114889125733978e-05,
"loss": 2.1442,
"step": 4350
},
{
"epoch": 1.23,
"learning_rate": 1.601105299065436e-05,
"loss": 2.1158,
"step": 4360
},
{
"epoch": 1.24,
"learning_rate": 1.5907394645222164e-05,
"loss": 2.1649,
"step": 4370
},
{
"epoch": 1.24,
"learning_rate": 1.5803916139664247e-05,
"loss": 2.0916,
"step": 4380
},
{
"epoch": 1.24,
"learning_rate": 1.5700619520650484e-05,
"loss": 2.1236,
"step": 4390
},
{
"epoch": 1.25,
"learning_rate": 1.5597506831253276e-05,
"loss": 2.1343,
"step": 4400
},
{
"epoch": 1.25,
"learning_rate": 1.5494580110907114e-05,
"loss": 2.1721,
"step": 4410
},
{
"epoch": 1.25,
"learning_rate": 1.5391841395368294e-05,
"loss": 2.1265,
"step": 4420
},
{
"epoch": 1.25,
"learning_rate": 1.5289292716674584e-05,
"loss": 2.1174,
"step": 4430
},
{
"epoch": 1.26,
"learning_rate": 1.5186936103105115e-05,
"loss": 2.1656,
"step": 4440
},
{
"epoch": 1.26,
"learning_rate": 1.5084773579140185e-05,
"loss": 2.1077,
"step": 4450
},
{
"epoch": 1.26,
"learning_rate": 1.4982807165421273e-05,
"loss": 2.1238,
"step": 4460
},
{
"epoch": 1.27,
"learning_rate": 1.4881038878711028e-05,
"loss": 2.1414,
"step": 4470
},
{
"epoch": 1.27,
"learning_rate": 1.4779470731853449e-05,
"loss": 2.1499,
"step": 4480
},
{
"epoch": 1.27,
"learning_rate": 1.4678104733733992e-05,
"loss": 2.1346,
"step": 4490
},
{
"epoch": 1.27,
"learning_rate": 1.457694288923988e-05,
"loss": 2.1281,
"step": 4500
},
{
"epoch": 1.28,
"learning_rate": 1.4475987199220465e-05,
"loss": 2.1302,
"step": 4510
},
{
"epoch": 1.28,
"learning_rate": 1.4375239660447599e-05,
"loss": 2.137,
"step": 4520
},
{
"epoch": 1.28,
"learning_rate": 1.4274702265576206e-05,
"loss": 2.1567,
"step": 4530
},
{
"epoch": 1.29,
"learning_rate": 1.4174377003104816e-05,
"loss": 2.1252,
"step": 4540
},
{
"epoch": 1.29,
"learning_rate": 1.4074265857336277e-05,
"loss": 2.1205,
"step": 4550
},
{
"epoch": 1.29,
"learning_rate": 1.3974370808338455e-05,
"loss": 2.1126,
"step": 4560
},
{
"epoch": 1.29,
"learning_rate": 1.387469383190515e-05,
"loss": 2.1677,
"step": 4570
},
{
"epoch": 1.3,
"learning_rate": 1.3775236899516933e-05,
"loss": 2.1687,
"step": 4580
},
{
"epoch": 1.3,
"learning_rate": 1.36760019783022e-05,
"loss": 2.105,
"step": 4590
},
{
"epoch": 1.3,
"learning_rate": 1.3576991030998254e-05,
"loss": 2.1517,
"step": 4600
},
{
"epoch": 1.31,
"learning_rate": 1.3478206015912509e-05,
"loss": 2.1052,
"step": 4610
},
{
"epoch": 1.31,
"learning_rate": 1.3379648886883694e-05,
"loss": 2.1441,
"step": 4620
},
{
"epoch": 1.31,
"learning_rate": 1.3281321593243296e-05,
"loss": 2.1338,
"step": 4630
},
{
"epoch": 1.31,
"learning_rate": 1.3183226079776911e-05,
"loss": 2.1139,
"step": 4640
},
{
"epoch": 1.32,
"learning_rate": 1.308536428668587e-05,
"loss": 2.1043,
"step": 4650
},
{
"epoch": 1.32,
"learning_rate": 1.2987738149548781e-05,
"loss": 2.1456,
"step": 4660
},
{
"epoch": 1.32,
"learning_rate": 1.2890349599283308e-05,
"loss": 2.1613,
"step": 4670
},
{
"epoch": 1.33,
"learning_rate": 1.2793200562107938e-05,
"loss": 2.1204,
"step": 4680
},
{
"epoch": 1.33,
"learning_rate": 1.2696292959503922e-05,
"loss": 2.1378,
"step": 4690
},
{
"epoch": 1.33,
"learning_rate": 1.2599628708177253e-05,
"loss": 2.1361,
"step": 4700
},
{
"epoch": 1.33,
"learning_rate": 1.250320972002072e-05,
"loss": 2.1881,
"step": 4710
},
{
"epoch": 1.34,
"learning_rate": 1.2407037902076168e-05,
"loss": 2.0997,
"step": 4720
},
{
"epoch": 1.34,
"learning_rate": 1.23111151564967e-05,
"loss": 2.1177,
"step": 4730
},
{
"epoch": 1.34,
"learning_rate": 1.2215443380509125e-05,
"loss": 2.1278,
"step": 4740
},
{
"epoch": 1.34,
"learning_rate": 1.2120024466376381e-05,
"loss": 2.1691,
"step": 4750
},
{
"epoch": 1.35,
"learning_rate": 1.2024860301360131e-05,
"loss": 2.1266,
"step": 4760
},
{
"epoch": 1.35,
"learning_rate": 1.1929952767683426e-05,
"loss": 2.1329,
"step": 4770
},
{
"epoch": 1.35,
"learning_rate": 1.1835303742493508e-05,
"loss": 2.1298,
"step": 4780
},
{
"epoch": 1.36,
"learning_rate": 1.1740915097824629e-05,
"loss": 2.136,
"step": 4790
},
{
"epoch": 1.36,
"learning_rate": 1.1646788700561085e-05,
"loss": 2.095,
"step": 4800
},
{
"epoch": 1.36,
"learning_rate": 1.1552926412400225e-05,
"loss": 2.1709,
"step": 4810
},
{
"epoch": 1.36,
"learning_rate": 1.1459330089815699e-05,
"loss": 2.1401,
"step": 4820
},
{
"epoch": 1.37,
"learning_rate": 1.136600158402068e-05,
"loss": 2.1297,
"step": 4830
},
{
"epoch": 1.37,
"learning_rate": 1.1272942740931286e-05,
"loss": 2.1329,
"step": 4840
},
{
"epoch": 1.37,
"learning_rate": 1.1180155401130044e-05,
"loss": 2.1276,
"step": 4850
},
{
"epoch": 1.38,
"learning_rate": 1.1087641399829527e-05,
"loss": 2.1066,
"step": 4860
},
{
"epoch": 1.38,
"learning_rate": 1.0995402566835997e-05,
"loss": 2.1604,
"step": 4870
},
{
"epoch": 1.38,
"learning_rate": 1.090344072651328e-05,
"loss": 2.1724,
"step": 4880
},
{
"epoch": 1.38,
"learning_rate": 1.0811757697746614e-05,
"loss": 2.1361,
"step": 4890
},
{
"epoch": 1.39,
"learning_rate": 1.0720355293906741e-05,
"loss": 2.0965,
"step": 4900
},
{
"epoch": 1.39,
"learning_rate": 1.0629235322813982e-05,
"loss": 2.1209,
"step": 4910
},
{
"epoch": 1.39,
"learning_rate": 1.0538399586702507e-05,
"loss": 2.149,
"step": 4920
},
{
"epoch": 1.4,
"learning_rate": 1.0447849882184715e-05,
"loss": 2.1216,
"step": 4930
},
{
"epoch": 1.4,
"learning_rate": 1.035758800021564e-05,
"loss": 2.1124,
"step": 4940
},
{
"epoch": 1.4,
"learning_rate": 1.0267615726057592e-05,
"loss": 2.1417,
"step": 4950
},
{
"epoch": 1.4,
"learning_rate": 1.0177934839244787e-05,
"loss": 2.1586,
"step": 4960
},
{
"epoch": 1.41,
"learning_rate": 1.0088547113548211e-05,
"loss": 2.1126,
"step": 4970
},
{
"epoch": 1.41,
"learning_rate": 9.99945431694048e-06,
"loss": 2.1373,
"step": 4980
},
{
"epoch": 1.41,
"learning_rate": 9.910658211560902e-06,
"loss": 2.1604,
"step": 4990
},
{
"epoch": 1.42,
"learning_rate": 9.822160553680617e-06,
"loss": 2.1359,
"step": 5000
},
{
"epoch": 1.42,
"learning_rate": 9.733963093667889e-06,
"loss": 2.0978,
"step": 5010
},
{
"epoch": 1.42,
"learning_rate": 9.646067575953419e-06,
"loss": 2.1111,
"step": 5020
},
{
"epoch": 1.42,
"learning_rate": 9.558475738995923e-06,
"loss": 2.1184,
"step": 5030
},
{
"epoch": 1.43,
"learning_rate": 9.471189315247669e-06,
"loss": 2.1054,
"step": 5040
},
{
"epoch": 1.43,
"learning_rate": 9.384210031120288e-06,
"loss": 2.1198,
"step": 5050
},
{
"epoch": 1.43,
"learning_rate": 9.297539606950564e-06,
"loss": 2.1512,
"step": 5060
},
{
"epoch": 1.44,
"learning_rate": 9.21117975696644e-06,
"loss": 2.1449,
"step": 5070
},
{
"epoch": 1.44,
"learning_rate": 9.1251321892531e-06,
"loss": 2.1285,
"step": 5080
},
{
"epoch": 1.44,
"learning_rate": 9.039398605719215e-06,
"loss": 2.1423,
"step": 5090
},
{
"epoch": 1.44,
"learning_rate": 8.953980702063233e-06,
"loss": 2.137,
"step": 5100
},
{
"epoch": 1.45,
"learning_rate": 8.868880167739887e-06,
"loss": 2.1535,
"step": 5110
},
{
"epoch": 1.45,
"learning_rate": 8.784098685926734e-06,
"loss": 2.1345,
"step": 5120
},
{
"epoch": 1.45,
"learning_rate": 8.699637933490917e-06,
"loss": 2.1687,
"step": 5130
},
{
"epoch": 1.46,
"learning_rate": 8.615499580955953e-06,
"loss": 2.1168,
"step": 5140
},
{
"epoch": 1.46,
"learning_rate": 8.531685292468693e-06,
"loss": 2.1382,
"step": 5150
},
{
"epoch": 1.46,
"learning_rate": 8.448196725766471e-06,
"loss": 2.1593,
"step": 5160
},
{
"epoch": 1.46,
"learning_rate": 8.365035532144217e-06,
"loss": 2.1451,
"step": 5170
},
{
"epoch": 1.47,
"learning_rate": 8.282203356421891e-06,
"loss": 2.1412,
"step": 5180
},
{
"epoch": 1.47,
"learning_rate": 8.199701836911877e-06,
"loss": 2.1484,
"step": 5190
},
{
"epoch": 1.47,
"learning_rate": 8.117532605386632e-06,
"loss": 2.1001,
"step": 5200
},
{
"epoch": 1.48,
"learning_rate": 8.035697287046376e-06,
"loss": 2.152,
"step": 5210
},
{
"epoch": 1.48,
"learning_rate": 7.954197500486968e-06,
"loss": 2.1376,
"step": 5220
},
{
"epoch": 1.48,
"learning_rate": 7.873034857667877e-06,
"loss": 2.1336,
"step": 5230
},
{
"epoch": 1.48,
"learning_rate": 7.792210963880328e-06,
"loss": 2.1333,
"step": 5240
},
{
"epoch": 1.49,
"learning_rate": 7.711727417715509e-06,
"loss": 2.1019,
"step": 5250
},
{
"epoch": 1.49,
"learning_rate": 7.631585811032999e-06,
"loss": 2.1241,
"step": 5260
},
{
"epoch": 1.49,
"learning_rate": 7.551787728929236e-06,
"loss": 2.0995,
"step": 5270
},
{
"epoch": 1.49,
"learning_rate": 7.472334749706214e-06,
"loss": 2.1497,
"step": 5280
},
{
"epoch": 1.5,
"learning_rate": 7.393228444840228e-06,
"loss": 2.1005,
"step": 5290
},
{
"epoch": 1.5,
"learning_rate": 7.314470378950802e-06,
"loss": 2.1144,
"step": 5300
},
{
"epoch": 1.5,
"learning_rate": 7.23606210976975e-06,
"loss": 2.1308,
"step": 5310
},
{
"epoch": 1.51,
"learning_rate": 7.158005188110381e-06,
"loss": 2.144,
"step": 5320
},
{
"epoch": 1.51,
"learning_rate": 7.080301157836783e-06,
"loss": 2.1136,
"step": 5330
},
{
"epoch": 1.51,
"learning_rate": 7.002951555833334e-06,
"loss": 2.1023,
"step": 5340
},
{
"epoch": 1.51,
"learning_rate": 6.925957911974262e-06,
"loss": 2.0962,
"step": 5350
},
{
"epoch": 1.52,
"learning_rate": 6.849321749093432e-06,
"loss": 2.1435,
"step": 5360
},
{
"epoch": 1.52,
"learning_rate": 6.773044582954172e-06,
"loss": 2.1234,
"step": 5370
},
{
"epoch": 1.52,
"learning_rate": 6.697127922219357e-06,
"loss": 2.1294,
"step": 5380
},
{
"epoch": 1.53,
"learning_rate": 6.621573268421505e-06,
"loss": 2.148,
"step": 5390
},
{
"epoch": 1.53,
"learning_rate": 6.546382115933117e-06,
"loss": 2.1679,
"step": 5400
},
{
"epoch": 1.53,
"learning_rate": 6.47155595193713e-06,
"loss": 2.1512,
"step": 5410
},
{
"epoch": 1.53,
"learning_rate": 6.397096256397456e-06,
"loss": 2.1235,
"step": 5420
},
{
"epoch": 1.54,
"learning_rate": 6.323004502029767e-06,
"loss": 2.1097,
"step": 5430
},
{
"epoch": 1.54,
"learning_rate": 6.249282154272315e-06,
"loss": 2.1027,
"step": 5440
},
{
"epoch": 1.54,
"learning_rate": 6.175930671256991e-06,
"loss": 2.1232,
"step": 5450
},
{
"epoch": 1.55,
"learning_rate": 6.102951503780449e-06,
"loss": 2.1662,
"step": 5460
},
{
"epoch": 1.55,
"learning_rate": 6.030346095275438e-06,
"loss": 2.1194,
"step": 5470
},
{
"epoch": 1.55,
"learning_rate": 5.958115881782228e-06,
"loss": 2.1317,
"step": 5480
},
{
"epoch": 1.55,
"learning_rate": 5.886262291920244e-06,
"loss": 2.1645,
"step": 5490
},
{
"epoch": 1.56,
"learning_rate": 5.81478674685976e-06,
"loss": 2.1165,
"step": 5500
},
{
"epoch": 1.56,
"learning_rate": 5.7436906602938435e-06,
"loss": 2.1565,
"step": 5510
},
{
"epoch": 1.56,
"learning_rate": 5.672975438410344e-06,
"loss": 2.1405,
"step": 5520
},
{
"epoch": 1.57,
"learning_rate": 5.602642479864129e-06,
"loss": 2.1334,
"step": 5530
},
{
"epoch": 1.57,
"learning_rate": 5.532693175749373e-06,
"loss": 2.0949,
"step": 5540
},
{
"epoch": 1.57,
"learning_rate": 5.463128909572077e-06,
"loss": 2.0991,
"step": 5550
},
{
"epoch": 1.57,
"learning_rate": 5.393951057222685e-06,
"loss": 2.1313,
"step": 5560
},
{
"epoch": 1.58,
"learning_rate": 5.325160986948899e-06,
"loss": 2.1146,
"step": 5570
},
{
"epoch": 1.58,
"learning_rate": 5.256760059328572e-06,
"loss": 2.0979,
"step": 5580
},
{
"epoch": 1.58,
"learning_rate": 5.188749627242851e-06,
"loss": 2.1453,
"step": 5590
},
{
"epoch": 1.59,
"learning_rate": 5.121131035849361e-06,
"loss": 2.1265,
"step": 5600
},
{
"epoch": 1.59,
"learning_rate": 5.053905622555657e-06,
"loss": 2.1417,
"step": 5610
},
{
"epoch": 1.59,
"learning_rate": 4.987074716992724e-06,
"loss": 2.1433,
"step": 5620
},
{
"epoch": 1.59,
"learning_rate": 4.920639640988697e-06,
"loss": 2.1306,
"step": 5630
},
{
"epoch": 1.6,
"learning_rate": 4.854601708542747e-06,
"loss": 2.1415,
"step": 5640
},
{
"epoch": 1.6,
"learning_rate": 4.7889622257990205e-06,
"loss": 2.1384,
"step": 5650
},
{
"epoch": 1.6,
"learning_rate": 4.7237224910208935e-06,
"loss": 2.1241,
"step": 5660
},
{
"epoch": 1.61,
"learning_rate": 4.658883794565208e-06,
"loss": 2.1171,
"step": 5670
},
{
"epoch": 1.61,
"learning_rate": 4.594447418856823e-06,
"loss": 2.1288,
"step": 5680
},
{
"epoch": 1.61,
"learning_rate": 4.530414638363201e-06,
"loss": 2.1165,
"step": 5690
},
{
"epoch": 1.61,
"learning_rate": 4.466786719569221e-06,
"loss": 2.1283,
"step": 5700
},
{
"epoch": 1.62,
"learning_rate": 4.403564920952116e-06,
"loss": 2.128,
"step": 5710
},
{
"epoch": 1.62,
"learning_rate": 4.3407504929566166e-06,
"loss": 2.1323,
"step": 5720
},
{
"epoch": 1.62,
"learning_rate": 4.278344677970167e-06,
"loss": 2.1326,
"step": 5730
},
{
"epoch": 1.63,
"learning_rate": 4.216348710298401e-06,
"loss": 2.1121,
"step": 5740
},
{
"epoch": 1.63,
"learning_rate": 4.154763816140686e-06,
"loss": 2.1193,
"step": 5750
},
{
"epoch": 1.63,
"learning_rate": 4.093591213565914e-06,
"loss": 2.145,
"step": 5760
},
{
"epoch": 1.63,
"learning_rate": 4.032832112488369e-06,
"loss": 2.1464,
"step": 5770
},
{
"epoch": 1.64,
"learning_rate": 3.972487714643822e-06,
"loss": 2.1353,
"step": 5780
},
{
"epoch": 1.64,
"learning_rate": 3.9125592135657475e-06,
"loss": 2.1069,
"step": 5790
},
{
"epoch": 1.64,
"learning_rate": 3.853047794561743e-06,
"loss": 2.1326,
"step": 5800
},
{
"epoch": 1.64,
"learning_rate": 3.7939546346900417e-06,
"loss": 2.1025,
"step": 5810
},
{
"epoch": 1.65,
"learning_rate": 3.7352809027362796e-06,
"loss": 2.0909,
"step": 5820
},
{
"epoch": 1.65,
"learning_rate": 3.6770277591903462e-06,
"loss": 2.0791,
"step": 5830
},
{
"epoch": 1.65,
"learning_rate": 3.619196356223439e-06,
"loss": 2.114,
"step": 5840
},
{
"epoch": 1.66,
"learning_rate": 3.5617878376652785e-06,
"loss": 2.1278,
"step": 5850
},
{
"epoch": 1.66,
"learning_rate": 3.504803338981472e-06,
"loss": 2.1344,
"step": 5860
},
{
"epoch": 1.66,
"learning_rate": 3.448243987251093e-06,
"loss": 2.1379,
"step": 5870
},
{
"epoch": 1.66,
"learning_rate": 3.3921109011443354e-06,
"loss": 2.1119,
"step": 5880
},
{
"epoch": 1.67,
"learning_rate": 3.336405190900438e-06,
"loss": 2.1633,
"step": 5890
},
{
"epoch": 1.67,
"learning_rate": 3.2811279583056825e-06,
"loss": 2.1225,
"step": 5900
},
{
"epoch": 1.67,
"learning_rate": 3.2262802966716383e-06,
"loss": 2.1153,
"step": 5910
},
{
"epoch": 1.68,
"learning_rate": 3.171863290813509e-06,
"loss": 2.1923,
"step": 5920
},
{
"epoch": 1.68,
"learning_rate": 3.117878017028697e-06,
"loss": 2.1015,
"step": 5930
},
{
"epoch": 1.68,
"learning_rate": 3.064325543075494e-06,
"loss": 2.1306,
"step": 5940
},
{
"epoch": 1.68,
"learning_rate": 3.0112069281519994e-06,
"loss": 2.1154,
"step": 5950
},
{
"epoch": 1.69,
"learning_rate": 2.9585232228751204e-06,
"loss": 2.1123,
"step": 5960
},
{
"epoch": 1.69,
"learning_rate": 2.906275469259839e-06,
"loss": 2.1266,
"step": 5970
},
{
"epoch": 1.69,
"learning_rate": 2.8544647006985724e-06,
"loss": 2.1457,
"step": 5980
},
{
"epoch": 1.7,
"learning_rate": 2.803091941940747e-06,
"loss": 2.1493,
"step": 5990
},
{
"epoch": 1.7,
"learning_rate": 2.752158209072522e-06,
"loss": 2.1412,
"step": 6000
}
],
"max_steps": 7064,
"num_train_epochs": 2,
"total_flos": 1.5290810330544865e+18,
"trial_name": null,
"trial_params": null
}