svjack's picture
Upload 55 files
50b1153 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.6957862281603289,
"eval_steps": 500,
"global_step": 3300,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.9999979150098865e-05,
"loss": 4.0418,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 4.999989444743508e-05,
"loss": 3.8361,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 4.999974458911041e-05,
"loss": 3.8737,
"step": 15
},
{
"epoch": 0.01,
"learning_rate": 4.9999529575515425e-05,
"loss": 3.806,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 4.9999249407210515e-05,
"loss": 3.4125,
"step": 25
},
{
"epoch": 0.02,
"learning_rate": 4.9998904084925845e-05,
"loss": 3.6489,
"step": 30
},
{
"epoch": 0.02,
"learning_rate": 4.999849360956141e-05,
"loss": 3.4719,
"step": 35
},
{
"epoch": 0.02,
"learning_rate": 4.999811831976564e-05,
"loss": 3.5265,
"step": 40
},
{
"epoch": 0.02,
"learning_rate": 4.999759057166764e-05,
"loss": 3.6889,
"step": 45
},
{
"epoch": 0.03,
"learning_rate": 4.9996997673913204e-05,
"loss": 3.5104,
"step": 50
},
{
"epoch": 0.03,
"learning_rate": 4.999633962804756e-05,
"loss": 3.5778,
"step": 55
},
{
"epoch": 0.03,
"learning_rate": 4.999561643578575e-05,
"loss": 3.3353,
"step": 60
},
{
"epoch": 0.03,
"learning_rate": 4.999482809901257e-05,
"loss": 3.3511,
"step": 65
},
{
"epoch": 0.04,
"learning_rate": 4.9993974619782625e-05,
"loss": 3.6355,
"step": 70
},
{
"epoch": 0.04,
"learning_rate": 4.999305600032027e-05,
"loss": 3.3915,
"step": 75
},
{
"epoch": 0.04,
"learning_rate": 4.999207224301965e-05,
"loss": 3.4926,
"step": 80
},
{
"epoch": 0.04,
"learning_rate": 4.999102335044467e-05,
"loss": 3.4918,
"step": 85
},
{
"epoch": 0.05,
"learning_rate": 4.9989909325328996e-05,
"loss": 3.3984,
"step": 90
},
{
"epoch": 0.05,
"learning_rate": 4.998873017057605e-05,
"loss": 3.4989,
"step": 95
},
{
"epoch": 0.05,
"learning_rate": 4.998748588925897e-05,
"loss": 3.6368,
"step": 100
},
{
"epoch": 0.05,
"learning_rate": 4.9986176484620665e-05,
"loss": 3.4412,
"step": 105
},
{
"epoch": 0.06,
"learning_rate": 4.998480196007375e-05,
"loss": 3.4708,
"step": 110
},
{
"epoch": 0.06,
"learning_rate": 4.9983362319200554e-05,
"loss": 3.37,
"step": 115
},
{
"epoch": 0.06,
"learning_rate": 4.998185756575313e-05,
"loss": 3.4524,
"step": 120
},
{
"epoch": 0.06,
"learning_rate": 4.9980287703653225e-05,
"loss": 3.3863,
"step": 125
},
{
"epoch": 0.07,
"learning_rate": 4.997865273699227e-05,
"loss": 3.4414,
"step": 130
},
{
"epoch": 0.07,
"learning_rate": 4.997729789123732e-05,
"loss": 3.3335,
"step": 135
},
{
"epoch": 0.07,
"learning_rate": 4.997554574721849e-05,
"loss": 3.231,
"step": 140
},
{
"epoch": 0.07,
"learning_rate": 4.997372851099727e-05,
"loss": 3.5962,
"step": 145
},
{
"epoch": 0.08,
"learning_rate": 4.997184618730983e-05,
"loss": 3.2795,
"step": 150
},
{
"epoch": 0.08,
"learning_rate": 4.996989878106192e-05,
"loss": 3.5384,
"step": 155
},
{
"epoch": 0.08,
"learning_rate": 4.996788629732897e-05,
"loss": 3.3274,
"step": 160
},
{
"epoch": 0.08,
"learning_rate": 4.996580874135597e-05,
"loss": 3.2292,
"step": 165
},
{
"epoch": 0.09,
"learning_rate": 4.996366611855753e-05,
"loss": 3.4829,
"step": 170
},
{
"epoch": 0.09,
"learning_rate": 4.996145843451785e-05,
"loss": 3.1998,
"step": 175
},
{
"epoch": 0.09,
"learning_rate": 4.9959185694990655e-05,
"loss": 3.3757,
"step": 180
},
{
"epoch": 0.1,
"learning_rate": 4.995684790589926e-05,
"loss": 3.1001,
"step": 185
},
{
"epoch": 0.1,
"learning_rate": 4.99544450733365e-05,
"loss": 3.2301,
"step": 190
},
{
"epoch": 0.1,
"learning_rate": 4.995197720356471e-05,
"loss": 3.3187,
"step": 195
},
{
"epoch": 0.1,
"learning_rate": 4.994944430301576e-05,
"loss": 3.527,
"step": 200
},
{
"epoch": 0.11,
"learning_rate": 4.994684637829098e-05,
"loss": 3.3808,
"step": 205
},
{
"epoch": 0.11,
"learning_rate": 4.994418343616119e-05,
"loss": 3.2406,
"step": 210
},
{
"epoch": 0.11,
"learning_rate": 4.994145548356664e-05,
"loss": 3.3463,
"step": 215
},
{
"epoch": 0.11,
"learning_rate": 4.993866252761702e-05,
"loss": 3.2972,
"step": 220
},
{
"epoch": 0.12,
"learning_rate": 4.993580457559145e-05,
"loss": 3.4135,
"step": 225
},
{
"epoch": 0.12,
"learning_rate": 4.9932881634938424e-05,
"loss": 3.4025,
"step": 230
},
{
"epoch": 0.12,
"learning_rate": 4.992989371327581e-05,
"loss": 3.662,
"step": 235
},
{
"epoch": 0.12,
"learning_rate": 4.992684081839086e-05,
"loss": 3.4279,
"step": 240
},
{
"epoch": 0.13,
"learning_rate": 4.9923722958240124e-05,
"loss": 3.3075,
"step": 245
},
{
"epoch": 0.13,
"learning_rate": 4.992054014094951e-05,
"loss": 3.2826,
"step": 250
},
{
"epoch": 0.13,
"learning_rate": 4.991729237481417e-05,
"loss": 3.4992,
"step": 255
},
{
"epoch": 0.13,
"learning_rate": 4.991397966829856e-05,
"loss": 3.2421,
"step": 260
},
{
"epoch": 0.14,
"learning_rate": 4.991060203003639e-05,
"loss": 3.2615,
"step": 265
},
{
"epoch": 0.14,
"learning_rate": 4.9907159468830585e-05,
"loss": 3.3299,
"step": 270
},
{
"epoch": 0.14,
"learning_rate": 4.990365199365327e-05,
"loss": 3.3117,
"step": 275
},
{
"epoch": 0.14,
"learning_rate": 4.9900079613645754e-05,
"loss": 3.4729,
"step": 280
},
{
"epoch": 0.15,
"learning_rate": 4.9896442338118524e-05,
"loss": 3.2521,
"step": 285
},
{
"epoch": 0.15,
"learning_rate": 4.989274017655117e-05,
"loss": 3.261,
"step": 290
},
{
"epoch": 0.15,
"learning_rate": 4.9888973138592414e-05,
"loss": 3.4403,
"step": 295
},
{
"epoch": 0.15,
"learning_rate": 4.988514123406004e-05,
"loss": 3.3408,
"step": 300
},
{
"epoch": 0.16,
"learning_rate": 4.9881244472940914e-05,
"loss": 3.3208,
"step": 305
},
{
"epoch": 0.16,
"learning_rate": 4.9877282865390904e-05,
"loss": 3.6889,
"step": 310
},
{
"epoch": 0.16,
"learning_rate": 4.9873256421734915e-05,
"loss": 3.1537,
"step": 315
},
{
"epoch": 0.16,
"learning_rate": 4.986916515246681e-05,
"loss": 3.1837,
"step": 320
},
{
"epoch": 0.17,
"learning_rate": 4.986500906824941e-05,
"loss": 3.23,
"step": 325
},
{
"epoch": 0.17,
"learning_rate": 4.9860788179914456e-05,
"loss": 3.4343,
"step": 330
},
{
"epoch": 0.17,
"learning_rate": 4.985650249846258e-05,
"loss": 3.3317,
"step": 335
},
{
"epoch": 0.17,
"learning_rate": 4.9852152035063294e-05,
"loss": 3.155,
"step": 340
},
{
"epoch": 0.18,
"learning_rate": 4.984773680105493e-05,
"loss": 3.3527,
"step": 345
},
{
"epoch": 0.18,
"learning_rate": 4.984415798673684e-05,
"loss": 3.3783,
"step": 350
},
{
"epoch": 0.18,
"learning_rate": 4.9839626194743575e-05,
"loss": 3.2567,
"step": 355
},
{
"epoch": 0.18,
"learning_rate": 4.983502966478656e-05,
"loss": 3.1278,
"step": 360
},
{
"epoch": 0.19,
"learning_rate": 4.9830368408845425e-05,
"loss": 3.2344,
"step": 365
},
{
"epoch": 0.19,
"learning_rate": 4.982564243906852e-05,
"loss": 3.3458,
"step": 370
},
{
"epoch": 0.19,
"learning_rate": 4.9820851767772844e-05,
"loss": 3.1467,
"step": 375
},
{
"epoch": 0.2,
"learning_rate": 4.981599640744402e-05,
"loss": 3.217,
"step": 380
},
{
"epoch": 0.2,
"learning_rate": 4.981107637073628e-05,
"loss": 3.4011,
"step": 385
},
{
"epoch": 0.2,
"learning_rate": 4.98060916704724e-05,
"loss": 3.4098,
"step": 390
},
{
"epoch": 0.2,
"learning_rate": 4.9801042319643686e-05,
"loss": 3.3332,
"step": 395
},
{
"epoch": 0.21,
"learning_rate": 4.979592833140996e-05,
"loss": 3.4042,
"step": 400
},
{
"epoch": 0.21,
"learning_rate": 4.979074971909948e-05,
"loss": 3.3127,
"step": 405
},
{
"epoch": 0.21,
"learning_rate": 4.978550649620894e-05,
"loss": 3.1726,
"step": 410
},
{
"epoch": 0.21,
"learning_rate": 4.978019867640342e-05,
"loss": 3.3029,
"step": 415
},
{
"epoch": 0.22,
"learning_rate": 4.977482627351637e-05,
"loss": 3.3519,
"step": 420
},
{
"epoch": 0.22,
"learning_rate": 4.976938930154954e-05,
"loss": 3.2507,
"step": 425
},
{
"epoch": 0.22,
"learning_rate": 4.976388777467297e-05,
"loss": 3.1322,
"step": 430
},
{
"epoch": 0.22,
"learning_rate": 4.9758321707224954e-05,
"loss": 3.1306,
"step": 435
},
{
"epoch": 0.23,
"learning_rate": 4.975269111371197e-05,
"loss": 3.189,
"step": 440
},
{
"epoch": 0.23,
"learning_rate": 4.974699600880869e-05,
"loss": 3.2598,
"step": 445
},
{
"epoch": 0.23,
"learning_rate": 4.974123640735791e-05,
"loss": 3.3483,
"step": 450
},
{
"epoch": 0.23,
"learning_rate": 4.973541232437052e-05,
"loss": 3.4678,
"step": 455
},
{
"epoch": 0.24,
"learning_rate": 4.972952377502545e-05,
"loss": 2.9668,
"step": 460
},
{
"epoch": 0.24,
"learning_rate": 4.9723570774669687e-05,
"loss": 3.372,
"step": 465
},
{
"epoch": 0.24,
"learning_rate": 4.971755333881814e-05,
"loss": 3.2808,
"step": 470
},
{
"epoch": 0.24,
"learning_rate": 4.971147148315367e-05,
"loss": 3.2235,
"step": 475
},
{
"epoch": 0.25,
"learning_rate": 4.9705325223527057e-05,
"loss": 3.0441,
"step": 480
},
{
"epoch": 0.25,
"learning_rate": 4.96991145759569e-05,
"loss": 3.4724,
"step": 485
},
{
"epoch": 0.25,
"learning_rate": 4.969283955662963e-05,
"loss": 3.389,
"step": 490
},
{
"epoch": 0.25,
"learning_rate": 4.968650018189943e-05,
"loss": 3.3817,
"step": 495
},
{
"epoch": 0.26,
"learning_rate": 4.968009646828822e-05,
"loss": 3.2597,
"step": 500
},
{
"epoch": 0.26,
"learning_rate": 4.967362843248559e-05,
"loss": 3.253,
"step": 505
},
{
"epoch": 0.26,
"learning_rate": 4.966709609134878e-05,
"loss": 3.1903,
"step": 510
},
{
"epoch": 0.26,
"learning_rate": 4.9660499461902645e-05,
"loss": 3.1286,
"step": 515
},
{
"epoch": 0.27,
"learning_rate": 4.965383856133953e-05,
"loss": 3.292,
"step": 520
},
{
"epoch": 0.27,
"learning_rate": 4.964711340701935e-05,
"loss": 3.4217,
"step": 525
},
{
"epoch": 0.27,
"learning_rate": 4.964032401646944e-05,
"loss": 3.2587,
"step": 530
},
{
"epoch": 0.27,
"learning_rate": 4.963347040738457e-05,
"loss": 3.3325,
"step": 535
},
{
"epoch": 0.28,
"learning_rate": 4.962655259762687e-05,
"loss": 3.3475,
"step": 540
},
{
"epoch": 0.28,
"learning_rate": 4.961957060522581e-05,
"loss": 3.2882,
"step": 545
},
{
"epoch": 0.28,
"learning_rate": 4.961252444837809e-05,
"loss": 3.3631,
"step": 550
},
{
"epoch": 0.29,
"learning_rate": 4.9605414145447696e-05,
"loss": 3.2589,
"step": 555
},
{
"epoch": 0.29,
"learning_rate": 4.959823971496574e-05,
"loss": 3.4855,
"step": 560
},
{
"epoch": 0.29,
"learning_rate": 4.959100117563052e-05,
"loss": 3.1354,
"step": 565
},
{
"epoch": 0.29,
"learning_rate": 4.958369854630737e-05,
"loss": 3.3037,
"step": 570
},
{
"epoch": 0.3,
"learning_rate": 4.957633184602867e-05,
"loss": 3.1023,
"step": 575
},
{
"epoch": 0.3,
"learning_rate": 4.956890109399381e-05,
"loss": 3.1933,
"step": 580
},
{
"epoch": 0.3,
"learning_rate": 4.956140630956908e-05,
"loss": 3.3994,
"step": 585
},
{
"epoch": 0.3,
"learning_rate": 4.955384751228767e-05,
"loss": 3.1178,
"step": 590
},
{
"epoch": 0.31,
"learning_rate": 4.954622472184961e-05,
"loss": 3.2513,
"step": 595
},
{
"epoch": 0.31,
"learning_rate": 4.9538537958121686e-05,
"loss": 3.5002,
"step": 600
},
{
"epoch": 0.31,
"learning_rate": 4.953078724113744e-05,
"loss": 3.1867,
"step": 605
},
{
"epoch": 0.31,
"learning_rate": 4.9522972591097074e-05,
"loss": 3.3883,
"step": 610
},
{
"epoch": 0.32,
"learning_rate": 4.9515094028367425e-05,
"loss": 3.3708,
"step": 615
},
{
"epoch": 0.32,
"learning_rate": 4.950715157348191e-05,
"loss": 3.2425,
"step": 620
},
{
"epoch": 0.32,
"learning_rate": 4.9499145247140434e-05,
"loss": 3.1046,
"step": 625
},
{
"epoch": 0.32,
"learning_rate": 4.9491075070209385e-05,
"loss": 3.3236,
"step": 630
},
{
"epoch": 0.33,
"learning_rate": 4.9482941063721576e-05,
"loss": 3.4957,
"step": 635
},
{
"epoch": 0.33,
"learning_rate": 4.947474324887614e-05,
"loss": 3.2225,
"step": 640
},
{
"epoch": 0.33,
"learning_rate": 4.9466481647038554e-05,
"loss": 3.5314,
"step": 645
},
{
"epoch": 0.33,
"learning_rate": 4.9458156279740484e-05,
"loss": 3.1028,
"step": 650
},
{
"epoch": 0.34,
"learning_rate": 4.944976716867984e-05,
"loss": 3.2322,
"step": 655
},
{
"epoch": 0.34,
"learning_rate": 4.9441314335720616e-05,
"loss": 3.1332,
"step": 660
},
{
"epoch": 0.34,
"learning_rate": 4.943279780289292e-05,
"loss": 3.2028,
"step": 665
},
{
"epoch": 0.34,
"learning_rate": 4.9424217592392854e-05,
"loss": 3.2408,
"step": 670
},
{
"epoch": 0.35,
"learning_rate": 4.9415573726582465e-05,
"loss": 3.0121,
"step": 675
},
{
"epoch": 0.35,
"learning_rate": 4.940686622798974e-05,
"loss": 3.2182,
"step": 680
},
{
"epoch": 0.35,
"learning_rate": 4.93980951193085e-05,
"loss": 3.4794,
"step": 685
},
{
"epoch": 0.35,
"learning_rate": 4.93892604233983e-05,
"loss": 3.3226,
"step": 690
},
{
"epoch": 0.36,
"learning_rate": 4.938036216328448e-05,
"loss": 3.1917,
"step": 695
},
{
"epoch": 0.36,
"learning_rate": 4.9371400362158016e-05,
"loss": 3.2176,
"step": 700
},
{
"epoch": 0.36,
"learning_rate": 4.936237504337549e-05,
"loss": 2.933,
"step": 705
},
{
"epoch": 0.36,
"learning_rate": 4.935328623045902e-05,
"loss": 3.134,
"step": 710
},
{
"epoch": 0.37,
"learning_rate": 4.934413394709621e-05,
"loss": 2.983,
"step": 715
},
{
"epoch": 0.37,
"learning_rate": 4.933491821714009e-05,
"loss": 3.4337,
"step": 720
},
{
"epoch": 0.37,
"learning_rate": 4.9325639064609045e-05,
"loss": 3.4699,
"step": 725
},
{
"epoch": 0.38,
"learning_rate": 4.931629651368673e-05,
"loss": 3.4081,
"step": 730
},
{
"epoch": 0.38,
"learning_rate": 4.930689058872208e-05,
"loss": 3.2174,
"step": 735
},
{
"epoch": 0.38,
"learning_rate": 4.929742131422913e-05,
"loss": 3.4953,
"step": 740
},
{
"epoch": 0.38,
"learning_rate": 4.92878887148871e-05,
"loss": 3.217,
"step": 745
},
{
"epoch": 0.39,
"learning_rate": 4.927829281554017e-05,
"loss": 3.2689,
"step": 750
},
{
"epoch": 0.39,
"learning_rate": 4.926863364119756e-05,
"loss": 3.2994,
"step": 755
},
{
"epoch": 0.39,
"learning_rate": 4.9258911217033355e-05,
"loss": 3.3868,
"step": 760
},
{
"epoch": 0.39,
"learning_rate": 4.924912556838651e-05,
"loss": 3.3169,
"step": 765
},
{
"epoch": 0.4,
"learning_rate": 4.923927672076075e-05,
"loss": 3.0466,
"step": 770
},
{
"epoch": 0.4,
"learning_rate": 4.92293646998245e-05,
"loss": 3.0857,
"step": 775
},
{
"epoch": 0.4,
"learning_rate": 4.921938953141084e-05,
"loss": 3.5001,
"step": 780
},
{
"epoch": 0.4,
"learning_rate": 4.920935124151744e-05,
"loss": 3.1633,
"step": 785
},
{
"epoch": 0.41,
"learning_rate": 4.919924985630647e-05,
"loss": 3.4816,
"step": 790
},
{
"epoch": 0.41,
"learning_rate": 4.9189085402104516e-05,
"loss": 3.2032,
"step": 795
},
{
"epoch": 0.41,
"learning_rate": 4.917885790540258e-05,
"loss": 3.164,
"step": 800
},
{
"epoch": 0.41,
"learning_rate": 4.916856739285593e-05,
"loss": 3.3047,
"step": 805
},
{
"epoch": 0.42,
"learning_rate": 4.91582138912841e-05,
"loss": 3.2366,
"step": 810
},
{
"epoch": 0.42,
"learning_rate": 4.9147797427670766e-05,
"loss": 3.3105,
"step": 815
},
{
"epoch": 0.42,
"learning_rate": 4.91373180291637e-05,
"loss": 3.4871,
"step": 820
},
{
"epoch": 0.42,
"learning_rate": 4.912677572307473e-05,
"loss": 3.0442,
"step": 825
},
{
"epoch": 0.43,
"learning_rate": 4.911617053687957e-05,
"loss": 3.1934,
"step": 830
},
{
"epoch": 0.43,
"learning_rate": 4.910550249821787e-05,
"loss": 3.27,
"step": 835
},
{
"epoch": 0.43,
"learning_rate": 4.9094771634893075e-05,
"loss": 3.3252,
"step": 840
},
{
"epoch": 0.43,
"learning_rate": 4.908397797487237e-05,
"loss": 2.9852,
"step": 845
},
{
"epoch": 0.44,
"learning_rate": 4.907312154628658e-05,
"loss": 3.1331,
"step": 850
},
{
"epoch": 0.44,
"learning_rate": 4.906220237743014e-05,
"loss": 3.1444,
"step": 855
},
{
"epoch": 0.44,
"learning_rate": 4.905122049676101e-05,
"loss": 3.3702,
"step": 860
},
{
"epoch": 0.44,
"learning_rate": 4.904017593290056e-05,
"loss": 3.0531,
"step": 865
},
{
"epoch": 0.45,
"learning_rate": 4.902906871463355e-05,
"loss": 3.2007,
"step": 870
},
{
"epoch": 0.45,
"learning_rate": 4.901789887090804e-05,
"loss": 3.0837,
"step": 875
},
{
"epoch": 0.45,
"learning_rate": 4.900666643083528e-05,
"loss": 3.2268,
"step": 880
},
{
"epoch": 0.45,
"learning_rate": 4.8995371423689684e-05,
"loss": 2.9249,
"step": 885
},
{
"epoch": 0.46,
"learning_rate": 4.898401387890873e-05,
"loss": 3.2058,
"step": 890
},
{
"epoch": 0.46,
"learning_rate": 4.897259382609285e-05,
"loss": 3.0971,
"step": 895
},
{
"epoch": 0.46,
"learning_rate": 4.8961111295005446e-05,
"loss": 3.2227,
"step": 900
},
{
"epoch": 0.47,
"learning_rate": 4.894956631557269e-05,
"loss": 3.2907,
"step": 905
},
{
"epoch": 0.47,
"learning_rate": 4.893795891788357e-05,
"loss": 3.1382,
"step": 910
},
{
"epoch": 0.47,
"learning_rate": 4.892628913218969e-05,
"loss": 2.8996,
"step": 915
},
{
"epoch": 0.47,
"learning_rate": 4.8914556988905315e-05,
"loss": 2.9693,
"step": 920
},
{
"epoch": 0.48,
"learning_rate": 4.890276251860717e-05,
"loss": 3.0134,
"step": 925
},
{
"epoch": 0.48,
"learning_rate": 4.889090575203445e-05,
"loss": 2.9696,
"step": 930
},
{
"epoch": 0.48,
"learning_rate": 4.887898672008871e-05,
"loss": 3.3306,
"step": 935
},
{
"epoch": 0.48,
"learning_rate": 4.8867005453833774e-05,
"loss": 3.2199,
"step": 940
},
{
"epoch": 0.49,
"learning_rate": 4.885496198449567e-05,
"loss": 3.2795,
"step": 945
},
{
"epoch": 0.49,
"learning_rate": 4.8842856343462535e-05,
"loss": 3.3482,
"step": 950
},
{
"epoch": 0.49,
"learning_rate": 4.883068856228454e-05,
"loss": 3.322,
"step": 955
},
{
"epoch": 0.49,
"learning_rate": 4.881845867267382e-05,
"loss": 3.0443,
"step": 960
},
{
"epoch": 0.5,
"learning_rate": 4.8806166706504375e-05,
"loss": 3.1826,
"step": 965
},
{
"epoch": 0.5,
"learning_rate": 4.879381269581197e-05,
"loss": 3.3732,
"step": 970
},
{
"epoch": 0.5,
"learning_rate": 4.878139667279411e-05,
"loss": 3.3457,
"step": 975
},
{
"epoch": 0.5,
"learning_rate": 4.876891866980988e-05,
"loss": 3.182,
"step": 980
},
{
"epoch": 0.51,
"learning_rate": 4.875637871937994e-05,
"loss": 3.2416,
"step": 985
},
{
"epoch": 0.51,
"learning_rate": 4.874377685418635e-05,
"loss": 3.2411,
"step": 990
},
{
"epoch": 0.51,
"learning_rate": 4.873111310707259e-05,
"loss": 3.2836,
"step": 995
},
{
"epoch": 0.51,
"learning_rate": 4.8718387511043385e-05,
"loss": 3.1773,
"step": 1000
},
{
"epoch": 0.52,
"learning_rate": 4.870560009926465e-05,
"loss": 3.1723,
"step": 1005
},
{
"epoch": 0.52,
"learning_rate": 4.8692750905063424e-05,
"loss": 3.1535,
"step": 1010
},
{
"epoch": 0.52,
"learning_rate": 4.867983996192776e-05,
"loss": 2.977,
"step": 1015
},
{
"epoch": 0.52,
"learning_rate": 4.866686730350663e-05,
"loss": 3.1236,
"step": 1020
},
{
"epoch": 0.53,
"learning_rate": 4.865383296360987e-05,
"loss": 3.4016,
"step": 1025
},
{
"epoch": 0.53,
"learning_rate": 4.864073697620807e-05,
"loss": 3.1547,
"step": 1030
},
{
"epoch": 0.53,
"learning_rate": 4.862757937543248e-05,
"loss": 3.143,
"step": 1035
},
{
"epoch": 0.53,
"learning_rate": 4.861436019557492e-05,
"loss": 3.1989,
"step": 1040
},
{
"epoch": 0.54,
"learning_rate": 4.860107947108772e-05,
"loss": 3.0308,
"step": 1045
},
{
"epoch": 0.54,
"learning_rate": 4.858773723658361e-05,
"loss": 3.409,
"step": 1050
},
{
"epoch": 0.54,
"learning_rate": 4.8574333526835605e-05,
"loss": 3.2011,
"step": 1055
},
{
"epoch": 0.54,
"learning_rate": 4.856086837677697e-05,
"loss": 3.1954,
"step": 1060
},
{
"epoch": 0.55,
"learning_rate": 4.854734182150107e-05,
"loss": 3.1366,
"step": 1065
},
{
"epoch": 0.55,
"learning_rate": 4.853375389626133e-05,
"loss": 3.3789,
"step": 1070
},
{
"epoch": 0.55,
"learning_rate": 4.85201046364711e-05,
"loss": 3.1337,
"step": 1075
},
{
"epoch": 0.55,
"learning_rate": 4.85063940777036e-05,
"loss": 3.2278,
"step": 1080
},
{
"epoch": 0.56,
"learning_rate": 4.849262225569181e-05,
"loss": 3.1339,
"step": 1085
},
{
"epoch": 0.56,
"learning_rate": 4.847878920632834e-05,
"loss": 2.954,
"step": 1090
},
{
"epoch": 0.56,
"learning_rate": 4.846489496566542e-05,
"loss": 3.0878,
"step": 1095
},
{
"epoch": 0.57,
"learning_rate": 4.845093956991475e-05,
"loss": 3.2329,
"step": 1100
},
{
"epoch": 0.57,
"learning_rate": 4.843692305544738e-05,
"loss": 3.2245,
"step": 1105
},
{
"epoch": 0.57,
"learning_rate": 4.84228454587937e-05,
"loss": 3.2865,
"step": 1110
},
{
"epoch": 0.57,
"learning_rate": 4.8408706816643254e-05,
"loss": 3.1569,
"step": 1115
},
{
"epoch": 0.58,
"learning_rate": 4.83945071658447e-05,
"loss": 3.2279,
"step": 1120
},
{
"epoch": 0.58,
"learning_rate": 4.838024654340571e-05,
"loss": 3.0554,
"step": 1125
},
{
"epoch": 0.58,
"learning_rate": 4.836592498649283e-05,
"loss": 3.3924,
"step": 1130
},
{
"epoch": 0.58,
"learning_rate": 4.835154253243147e-05,
"loss": 3.0695,
"step": 1135
},
{
"epoch": 0.59,
"learning_rate": 4.8337099218705695e-05,
"loss": 3.2666,
"step": 1140
},
{
"epoch": 0.59,
"learning_rate": 4.832259508295822e-05,
"loss": 3.135,
"step": 1145
},
{
"epoch": 0.59,
"learning_rate": 4.830803016299027e-05,
"loss": 2.947,
"step": 1150
},
{
"epoch": 0.59,
"learning_rate": 4.829340449676147e-05,
"loss": 3.13,
"step": 1155
},
{
"epoch": 0.6,
"learning_rate": 4.82787181223898e-05,
"loss": 3.038,
"step": 1160
},
{
"epoch": 0.6,
"learning_rate": 4.826397107815142e-05,
"loss": 3.0761,
"step": 1165
},
{
"epoch": 0.6,
"learning_rate": 4.824916340248064e-05,
"loss": 3.1496,
"step": 1170
},
{
"epoch": 0.6,
"learning_rate": 4.823429513396977e-05,
"loss": 2.8758,
"step": 1175
},
{
"epoch": 0.61,
"learning_rate": 4.821936631136906e-05,
"loss": 3.218,
"step": 1180
},
{
"epoch": 0.61,
"learning_rate": 4.820437697358656e-05,
"loss": 3.0933,
"step": 1185
},
{
"epoch": 0.61,
"learning_rate": 4.8189327159688035e-05,
"loss": 3.2388,
"step": 1190
},
{
"epoch": 0.61,
"learning_rate": 4.817421690889689e-05,
"loss": 3.1914,
"step": 1195
},
{
"epoch": 0.62,
"learning_rate": 4.815904626059401e-05,
"loss": 3.2259,
"step": 1200
},
{
"epoch": 0.62,
"learning_rate": 4.814381525431771e-05,
"loss": 3.1867,
"step": 1205
},
{
"epoch": 0.62,
"learning_rate": 4.8128523929763626e-05,
"loss": 3.1134,
"step": 1210
},
{
"epoch": 0.62,
"learning_rate": 4.811317232678456e-05,
"loss": 3.3846,
"step": 1215
},
{
"epoch": 0.63,
"learning_rate": 4.8097760485390444e-05,
"loss": 3.2507,
"step": 1220
},
{
"epoch": 0.63,
"learning_rate": 4.8082288445748194e-05,
"loss": 3.0992,
"step": 1225
},
{
"epoch": 0.63,
"learning_rate": 4.806675624818162e-05,
"loss": 3.0715,
"step": 1230
},
{
"epoch": 0.63,
"learning_rate": 4.8051163933171316e-05,
"loss": 3.2028,
"step": 1235
},
{
"epoch": 0.64,
"learning_rate": 4.8035511541354547e-05,
"loss": 3.2714,
"step": 1240
},
{
"epoch": 0.64,
"learning_rate": 4.801979911352519e-05,
"loss": 3.1238,
"step": 1245
},
{
"epoch": 0.64,
"learning_rate": 4.800402669063353e-05,
"loss": 3.2508,
"step": 1250
},
{
"epoch": 0.64,
"learning_rate": 4.7988194313786275e-05,
"loss": 3.2359,
"step": 1255
},
{
"epoch": 0.65,
"learning_rate": 4.797230202424633e-05,
"loss": 2.8538,
"step": 1260
},
{
"epoch": 0.65,
"learning_rate": 4.7956349863432794e-05,
"loss": 3.1991,
"step": 1265
},
{
"epoch": 0.65,
"learning_rate": 4.794033787292078e-05,
"loss": 3.0467,
"step": 1270
},
{
"epoch": 0.66,
"learning_rate": 4.7924266094441346e-05,
"loss": 3.1088,
"step": 1275
},
{
"epoch": 0.66,
"learning_rate": 4.7908134569881344e-05,
"loss": 3.128,
"step": 1280
},
{
"epoch": 0.66,
"learning_rate": 4.789194334128338e-05,
"loss": 3.3195,
"step": 1285
},
{
"epoch": 0.66,
"learning_rate": 4.7875692450845624e-05,
"loss": 3.2049,
"step": 1290
},
{
"epoch": 0.67,
"learning_rate": 4.785938194092176e-05,
"loss": 3.1477,
"step": 1295
},
{
"epoch": 0.67,
"learning_rate": 4.7843011854020846e-05,
"loss": 3.2506,
"step": 1300
},
{
"epoch": 0.67,
"learning_rate": 4.7826582232807213e-05,
"loss": 3.1822,
"step": 1305
},
{
"epoch": 0.67,
"learning_rate": 4.7810093120100374e-05,
"loss": 2.9576,
"step": 1310
},
{
"epoch": 0.68,
"learning_rate": 4.779354455887485e-05,
"loss": 3.2086,
"step": 1315
},
{
"epoch": 0.68,
"learning_rate": 4.777693659226013e-05,
"loss": 3.2935,
"step": 1320
},
{
"epoch": 0.68,
"learning_rate": 4.776026926354052e-05,
"loss": 2.9873,
"step": 1325
},
{
"epoch": 0.68,
"learning_rate": 4.774354261615503e-05,
"loss": 3.2671,
"step": 1330
},
{
"epoch": 0.69,
"learning_rate": 4.772675669369728e-05,
"loss": 3.0882,
"step": 1335
},
{
"epoch": 0.69,
"learning_rate": 4.770991153991537e-05,
"loss": 3.2727,
"step": 1340
},
{
"epoch": 0.69,
"learning_rate": 4.769300719871177e-05,
"loss": 2.9907,
"step": 1345
},
{
"epoch": 0.69,
"learning_rate": 4.7676043714143195e-05,
"loss": 3.1303,
"step": 1350
},
{
"epoch": 0.7,
"learning_rate": 4.765902113042053e-05,
"loss": 3.1412,
"step": 1355
},
{
"epoch": 0.7,
"learning_rate": 4.7641939491908674e-05,
"loss": 3.3585,
"step": 1360
},
{
"epoch": 0.7,
"learning_rate": 4.7624798843126404e-05,
"loss": 3.1958,
"step": 1365
},
{
"epoch": 0.7,
"learning_rate": 4.7607599228746355e-05,
"loss": 3.1581,
"step": 1370
},
{
"epoch": 0.71,
"learning_rate": 4.759034069359478e-05,
"loss": 3.224,
"step": 1375
},
{
"epoch": 0.71,
"learning_rate": 4.7573023282651536e-05,
"loss": 3.1031,
"step": 1380
},
{
"epoch": 0.71,
"learning_rate": 4.75556470410499e-05,
"loss": 3.0914,
"step": 1385
},
{
"epoch": 0.71,
"learning_rate": 4.753821201407648e-05,
"loss": 3.2474,
"step": 1390
},
{
"epoch": 0.72,
"learning_rate": 4.75207182471711e-05,
"loss": 3.2045,
"step": 1395
},
{
"epoch": 0.72,
"learning_rate": 4.7503165785926676e-05,
"loss": 2.9288,
"step": 1400
},
{
"epoch": 0.72,
"learning_rate": 4.748555467608909e-05,
"loss": 3.3084,
"step": 1405
},
{
"epoch": 0.72,
"learning_rate": 4.7467884963557064e-05,
"loss": 3.1761,
"step": 1410
},
{
"epoch": 0.73,
"learning_rate": 4.745015669438207e-05,
"loss": 3.0644,
"step": 1415
},
{
"epoch": 0.73,
"learning_rate": 4.7432369914768185e-05,
"loss": 3.0874,
"step": 1420
},
{
"epoch": 0.73,
"learning_rate": 4.7414524671071994e-05,
"loss": 3.1106,
"step": 1425
},
{
"epoch": 0.73,
"learning_rate": 4.739662100980242e-05,
"loss": 2.8364,
"step": 1430
},
{
"epoch": 0.74,
"learning_rate": 4.737865897762066e-05,
"loss": 3.0884,
"step": 1435
},
{
"epoch": 0.74,
"learning_rate": 4.736063862134004e-05,
"loss": 3.1134,
"step": 1440
},
{
"epoch": 0.74,
"learning_rate": 4.734255998792589e-05,
"loss": 3.3668,
"step": 1445
},
{
"epoch": 0.75,
"learning_rate": 4.732442312449541e-05,
"loss": 2.9823,
"step": 1450
},
{
"epoch": 0.75,
"learning_rate": 4.730622807831758e-05,
"loss": 3.0893,
"step": 1455
},
{
"epoch": 0.75,
"learning_rate": 4.728797489681301e-05,
"loss": 3.1091,
"step": 1460
},
{
"epoch": 0.75,
"learning_rate": 4.726966362755383e-05,
"loss": 2.9348,
"step": 1465
},
{
"epoch": 0.76,
"learning_rate": 4.725129431826355e-05,
"loss": 3.2506,
"step": 1470
},
{
"epoch": 0.76,
"learning_rate": 4.723286701681695e-05,
"loss": 2.9876,
"step": 1475
},
{
"epoch": 0.76,
"learning_rate": 4.7214381771239946e-05,
"loss": 2.9158,
"step": 1480
},
{
"epoch": 0.76,
"learning_rate": 4.719583862970949e-05,
"loss": 3.2035,
"step": 1485
},
{
"epoch": 0.77,
"learning_rate": 4.7177237640553414e-05,
"loss": 2.9031,
"step": 1490
},
{
"epoch": 0.77,
"learning_rate": 4.7158578852250304e-05,
"loss": 3.2779,
"step": 1495
},
{
"epoch": 0.77,
"learning_rate": 4.7139862313429385e-05,
"loss": 3.1219,
"step": 1500
},
{
"epoch": 0.77,
"learning_rate": 4.712108807287042e-05,
"loss": 3.1643,
"step": 1505
},
{
"epoch": 0.78,
"learning_rate": 4.7102256179503525e-05,
"loss": 3.1993,
"step": 1510
},
{
"epoch": 0.78,
"learning_rate": 4.708336668240908e-05,
"loss": 3.1773,
"step": 1515
},
{
"epoch": 0.78,
"learning_rate": 4.706441963081762e-05,
"loss": 2.9617,
"step": 1520
},
{
"epoch": 0.78,
"learning_rate": 4.7045415074109646e-05,
"loss": 3.2621,
"step": 1525
},
{
"epoch": 0.79,
"learning_rate": 4.702635306181554e-05,
"loss": 3.2206,
"step": 1530
},
{
"epoch": 0.79,
"learning_rate": 4.700723364361545e-05,
"loss": 3.3413,
"step": 1535
},
{
"epoch": 0.79,
"learning_rate": 4.6988056869339104e-05,
"loss": 3.3453,
"step": 1540
},
{
"epoch": 0.79,
"learning_rate": 4.696882278896574e-05,
"loss": 3.0481,
"step": 1545
},
{
"epoch": 0.8,
"learning_rate": 4.694953145262393e-05,
"loss": 3.2183,
"step": 1550
},
{
"epoch": 0.8,
"learning_rate": 4.693018291059147e-05,
"loss": 3.4063,
"step": 1555
},
{
"epoch": 0.8,
"learning_rate": 4.691077721329527e-05,
"loss": 3.1694,
"step": 1560
},
{
"epoch": 0.8,
"learning_rate": 4.6891314411311184e-05,
"loss": 3.0084,
"step": 1565
},
{
"epoch": 0.81,
"learning_rate": 4.6871794555363885e-05,
"loss": 3.318,
"step": 1570
},
{
"epoch": 0.81,
"learning_rate": 4.685221769632676e-05,
"loss": 3.3126,
"step": 1575
},
{
"epoch": 0.81,
"learning_rate": 4.683258388522175e-05,
"loss": 3.1085,
"step": 1580
},
{
"epoch": 0.81,
"learning_rate": 4.681289317321923e-05,
"loss": 3.0566,
"step": 1585
},
{
"epoch": 0.82,
"learning_rate": 4.679314561163787e-05,
"loss": 3.0642,
"step": 1590
},
{
"epoch": 0.82,
"learning_rate": 4.6773341251944496e-05,
"loss": 2.9335,
"step": 1595
},
{
"epoch": 0.82,
"learning_rate": 4.675348014575398e-05,
"loss": 3.1551,
"step": 1600
},
{
"epoch": 0.82,
"learning_rate": 4.673356234482908e-05,
"loss": 3.0211,
"step": 1605
},
{
"epoch": 0.83,
"learning_rate": 4.671358790108033e-05,
"loss": 3.28,
"step": 1610
},
{
"epoch": 0.83,
"learning_rate": 4.6693556866565835e-05,
"loss": 3.0621,
"step": 1615
},
{
"epoch": 0.83,
"learning_rate": 4.667346929349126e-05,
"loss": 3.1141,
"step": 1620
},
{
"epoch": 0.84,
"learning_rate": 4.6653325234209575e-05,
"loss": 3.1941,
"step": 1625
},
{
"epoch": 0.84,
"learning_rate": 4.663312474122099e-05,
"loss": 3.1874,
"step": 1630
},
{
"epoch": 0.84,
"learning_rate": 4.661286786717278e-05,
"loss": 3.2085,
"step": 1635
},
{
"epoch": 0.84,
"learning_rate": 4.659255466485918e-05,
"loss": 3.398,
"step": 1640
},
{
"epoch": 0.85,
"learning_rate": 4.65721851872212e-05,
"loss": 3.1911,
"step": 1645
},
{
"epoch": 0.85,
"learning_rate": 4.655175948734654e-05,
"loss": 3.2273,
"step": 1650
},
{
"epoch": 0.85,
"learning_rate": 4.653127761846943e-05,
"loss": 3.2084,
"step": 1655
},
{
"epoch": 0.85,
"learning_rate": 4.6510739633970485e-05,
"loss": 3.2789,
"step": 1660
},
{
"epoch": 0.86,
"learning_rate": 4.649014558737656e-05,
"loss": 3.1193,
"step": 1665
},
{
"epoch": 0.86,
"learning_rate": 4.646949553236064e-05,
"loss": 3.2083,
"step": 1670
},
{
"epoch": 0.86,
"learning_rate": 4.6448789522741665e-05,
"loss": 3.0306,
"step": 1675
},
{
"epoch": 0.86,
"learning_rate": 4.642802761248441e-05,
"loss": 3.0084,
"step": 1680
},
{
"epoch": 0.87,
"learning_rate": 4.6407209855699365e-05,
"loss": 3.2976,
"step": 1685
},
{
"epoch": 0.87,
"learning_rate": 4.6386336306642514e-05,
"loss": 3.3373,
"step": 1690
},
{
"epoch": 0.87,
"learning_rate": 4.63654070197153e-05,
"loss": 3.049,
"step": 1695
},
{
"epoch": 0.87,
"learning_rate": 4.634442204946442e-05,
"loss": 3.1669,
"step": 1700
},
{
"epoch": 0.88,
"learning_rate": 4.632338145058167e-05,
"loss": 3.1078,
"step": 1705
},
{
"epoch": 0.88,
"learning_rate": 4.630228527790386e-05,
"loss": 3.1724,
"step": 1710
},
{
"epoch": 0.88,
"learning_rate": 4.628113358641263e-05,
"loss": 3.2197,
"step": 1715
},
{
"epoch": 0.88,
"learning_rate": 4.625992643123429e-05,
"loss": 3.0047,
"step": 1720
},
{
"epoch": 0.89,
"learning_rate": 4.623866386763974e-05,
"loss": 2.9798,
"step": 1725
},
{
"epoch": 0.89,
"learning_rate": 4.621734595104427e-05,
"loss": 3.2065,
"step": 1730
},
{
"epoch": 0.89,
"learning_rate": 4.619597273700742e-05,
"loss": 3.0981,
"step": 1735
},
{
"epoch": 0.89,
"learning_rate": 4.6174544281232865e-05,
"loss": 2.7712,
"step": 1740
},
{
"epoch": 0.9,
"learning_rate": 4.615306063956826e-05,
"loss": 3.2469,
"step": 1745
},
{
"epoch": 0.9,
"learning_rate": 4.613152186800506e-05,
"loss": 3.3318,
"step": 1750
},
{
"epoch": 0.9,
"learning_rate": 4.610992802267842e-05,
"loss": 3.216,
"step": 1755
},
{
"epoch": 0.9,
"learning_rate": 4.608827915986704e-05,
"loss": 3.0298,
"step": 1760
},
{
"epoch": 0.91,
"learning_rate": 4.606657533599298e-05,
"loss": 3.1146,
"step": 1765
},
{
"epoch": 0.91,
"learning_rate": 4.604481660762158e-05,
"loss": 3.1621,
"step": 1770
},
{
"epoch": 0.91,
"learning_rate": 4.6023003031461234e-05,
"loss": 2.9654,
"step": 1775
},
{
"epoch": 0.91,
"learning_rate": 4.600113466436331e-05,
"loss": 3.3586,
"step": 1780
},
{
"epoch": 0.92,
"learning_rate": 4.597921156332196e-05,
"loss": 3.1212,
"step": 1785
},
{
"epoch": 0.92,
"learning_rate": 4.5957233785474e-05,
"loss": 2.8321,
"step": 1790
},
{
"epoch": 0.92,
"learning_rate": 4.5935201388098716e-05,
"loss": 2.9001,
"step": 1795
},
{
"epoch": 0.92,
"learning_rate": 4.5913114428617785e-05,
"loss": 3.248,
"step": 1800
},
{
"epoch": 0.93,
"learning_rate": 4.589097296459508e-05,
"loss": 2.8819,
"step": 1805
},
{
"epoch": 0.93,
"learning_rate": 4.5868777053736476e-05,
"loss": 3.0377,
"step": 1810
},
{
"epoch": 0.93,
"learning_rate": 4.584652675388982e-05,
"loss": 3.3161,
"step": 1815
},
{
"epoch": 0.94,
"learning_rate": 4.582422212304467e-05,
"loss": 3.1995,
"step": 1820
},
{
"epoch": 0.94,
"learning_rate": 4.580186321933217e-05,
"loss": 3.4196,
"step": 1825
},
{
"epoch": 0.94,
"learning_rate": 4.577945010102496e-05,
"loss": 3.35,
"step": 1830
},
{
"epoch": 0.94,
"learning_rate": 4.575698282653693e-05,
"loss": 3.2269,
"step": 1835
},
{
"epoch": 0.95,
"learning_rate": 4.573446145442314e-05,
"loss": 3.1207,
"step": 1840
},
{
"epoch": 0.95,
"learning_rate": 4.571188604337963e-05,
"loss": 3.3682,
"step": 1845
},
{
"epoch": 0.95,
"learning_rate": 4.568925665224327e-05,
"loss": 3.0202,
"step": 1850
},
{
"epoch": 0.95,
"learning_rate": 4.5666573339991646e-05,
"loss": 3.0227,
"step": 1855
},
{
"epoch": 0.96,
"learning_rate": 4.564383616574285e-05,
"loss": 3.1553,
"step": 1860
},
{
"epoch": 0.96,
"learning_rate": 4.562104518875535e-05,
"loss": 3.1961,
"step": 1865
},
{
"epoch": 0.96,
"learning_rate": 4.5598200468427854e-05,
"loss": 2.9751,
"step": 1870
},
{
"epoch": 0.96,
"learning_rate": 4.557530206429912e-05,
"loss": 2.9953,
"step": 1875
},
{
"epoch": 0.97,
"learning_rate": 4.555235003604782e-05,
"loss": 2.9119,
"step": 1880
},
{
"epoch": 0.97,
"learning_rate": 4.55293444434924e-05,
"loss": 3.1541,
"step": 1885
},
{
"epoch": 0.97,
"learning_rate": 4.5506285346590894e-05,
"loss": 2.9233,
"step": 1890
},
{
"epoch": 0.97,
"learning_rate": 4.5483172805440785e-05,
"loss": 3.3379,
"step": 1895
},
{
"epoch": 0.98,
"learning_rate": 4.546000688027884e-05,
"loss": 3.1169,
"step": 1900
},
{
"epoch": 0.98,
"learning_rate": 4.543678763148096e-05,
"loss": 3.3383,
"step": 1905
},
{
"epoch": 0.98,
"learning_rate": 4.5413515119562023e-05,
"loss": 3.1096,
"step": 1910
},
{
"epoch": 0.98,
"learning_rate": 4.539018940517572e-05,
"loss": 3.1755,
"step": 1915
},
{
"epoch": 0.99,
"learning_rate": 4.53668105491144e-05,
"loss": 3.1069,
"step": 1920
},
{
"epoch": 0.99,
"learning_rate": 4.534337861230891e-05,
"loss": 3.045,
"step": 1925
},
{
"epoch": 0.99,
"learning_rate": 4.5319893655828446e-05,
"loss": 2.975,
"step": 1930
},
{
"epoch": 0.99,
"learning_rate": 4.5296355740880385e-05,
"loss": 3.4444,
"step": 1935
},
{
"epoch": 1.0,
"learning_rate": 4.5272764928810115e-05,
"loss": 3.1052,
"step": 1940
},
{
"epoch": 1.0,
"learning_rate": 4.52491212811009e-05,
"loss": 3.0536,
"step": 1945
},
{
"epoch": 1.0,
"learning_rate": 4.522542485937369e-05,
"loss": 3.0542,
"step": 1950
},
{
"epoch": 1.0,
"learning_rate": 4.520167572538699e-05,
"loss": 3.158,
"step": 1955
},
{
"epoch": 1.01,
"learning_rate": 4.5177873941036683e-05,
"loss": 3.0392,
"step": 1960
},
{
"epoch": 1.01,
"learning_rate": 4.5154019568355876e-05,
"loss": 3.1487,
"step": 1965
},
{
"epoch": 1.01,
"learning_rate": 4.5130112669514726e-05,
"loss": 2.9061,
"step": 1970
},
{
"epoch": 1.01,
"learning_rate": 4.510615330682028e-05,
"loss": 2.9249,
"step": 1975
},
{
"epoch": 1.02,
"learning_rate": 4.5082141542716336e-05,
"loss": 3.0888,
"step": 1980
},
{
"epoch": 1.02,
"learning_rate": 4.505807743978324e-05,
"loss": 2.9417,
"step": 1985
},
{
"epoch": 1.02,
"learning_rate": 4.5033961060737774e-05,
"loss": 2.8967,
"step": 1990
},
{
"epoch": 1.03,
"learning_rate": 4.5009792468432946e-05,
"loss": 2.9344,
"step": 1995
},
{
"epoch": 1.03,
"learning_rate": 4.498557172585782e-05,
"loss": 3.1597,
"step": 2000
},
{
"epoch": 1.03,
"learning_rate": 4.496129889613744e-05,
"loss": 2.9188,
"step": 2005
},
{
"epoch": 1.03,
"learning_rate": 4.493697404253254e-05,
"loss": 2.7781,
"step": 2010
},
{
"epoch": 1.04,
"learning_rate": 4.4912597228439466e-05,
"loss": 3.0411,
"step": 2015
},
{
"epoch": 1.04,
"learning_rate": 4.488816851738999e-05,
"loss": 3.0727,
"step": 2020
},
{
"epoch": 1.04,
"learning_rate": 4.486368797305114e-05,
"loss": 3.0783,
"step": 2025
},
{
"epoch": 1.04,
"learning_rate": 4.483915565922501e-05,
"loss": 2.823,
"step": 2030
},
{
"epoch": 1.05,
"learning_rate": 4.481457163984864e-05,
"loss": 2.6414,
"step": 2035
},
{
"epoch": 1.05,
"learning_rate": 4.478993597899382e-05,
"loss": 2.9076,
"step": 2040
},
{
"epoch": 1.05,
"learning_rate": 4.476524874086695e-05,
"loss": 2.8118,
"step": 2045
},
{
"epoch": 1.05,
"learning_rate": 4.4740509989808815e-05,
"loss": 3.0352,
"step": 2050
},
{
"epoch": 1.06,
"learning_rate": 4.4715719790294474e-05,
"loss": 3.0765,
"step": 2055
},
{
"epoch": 1.06,
"learning_rate": 4.4690878206933086e-05,
"loss": 2.6218,
"step": 2060
},
{
"epoch": 1.06,
"learning_rate": 4.466598530446771e-05,
"loss": 3.0266,
"step": 2065
},
{
"epoch": 1.06,
"learning_rate": 4.464104114777515e-05,
"loss": 3.0852,
"step": 2070
},
{
"epoch": 1.07,
"learning_rate": 4.4616045801865814e-05,
"loss": 2.9536,
"step": 2075
},
{
"epoch": 1.07,
"learning_rate": 4.4590999331883505e-05,
"loss": 2.7272,
"step": 2080
},
{
"epoch": 1.07,
"learning_rate": 4.456590180310527e-05,
"loss": 3.0185,
"step": 2085
},
{
"epoch": 1.07,
"learning_rate": 4.454075328094123e-05,
"loss": 3.1669,
"step": 2090
},
{
"epoch": 1.08,
"learning_rate": 4.45155538309344e-05,
"loss": 2.7681,
"step": 2095
},
{
"epoch": 1.08,
"learning_rate": 4.449030351876053e-05,
"loss": 3.0262,
"step": 2100
},
{
"epoch": 1.08,
"learning_rate": 4.446500241022794e-05,
"loss": 2.7877,
"step": 2105
},
{
"epoch": 1.08,
"learning_rate": 4.443965057127731e-05,
"loss": 2.9973,
"step": 2110
},
{
"epoch": 1.09,
"learning_rate": 4.441424806798156e-05,
"loss": 2.9694,
"step": 2115
},
{
"epoch": 1.09,
"learning_rate": 4.438879496654565e-05,
"loss": 2.9082,
"step": 2120
},
{
"epoch": 1.09,
"learning_rate": 4.43632913333064e-05,
"loss": 3.1398,
"step": 2125
},
{
"epoch": 1.09,
"learning_rate": 4.4337737234732334e-05,
"loss": 2.9788,
"step": 2130
},
{
"epoch": 1.1,
"learning_rate": 4.431213273742351e-05,
"loss": 3.0885,
"step": 2135
},
{
"epoch": 1.1,
"learning_rate": 4.42864779081113e-05,
"loss": 3.1436,
"step": 2140
},
{
"epoch": 1.1,
"learning_rate": 4.4260772813658316e-05,
"loss": 2.9983,
"step": 2145
},
{
"epoch": 1.1,
"learning_rate": 4.423501752105812e-05,
"loss": 3.0249,
"step": 2150
},
{
"epoch": 1.11,
"learning_rate": 4.420921209743511e-05,
"loss": 2.8603,
"step": 2155
},
{
"epoch": 1.11,
"learning_rate": 4.4183356610044364e-05,
"loss": 2.8446,
"step": 2160
},
{
"epoch": 1.11,
"learning_rate": 4.415745112627141e-05,
"loss": 3.0299,
"step": 2165
},
{
"epoch": 1.12,
"learning_rate": 4.4131495713632104e-05,
"loss": 2.8172,
"step": 2170
},
{
"epoch": 1.12,
"learning_rate": 4.4105490439772404e-05,
"loss": 2.9729,
"step": 2175
},
{
"epoch": 1.12,
"learning_rate": 4.407943537246822e-05,
"loss": 2.9726,
"step": 2180
},
{
"epoch": 1.12,
"learning_rate": 4.4053330579625265e-05,
"loss": 2.8896,
"step": 2185
},
{
"epoch": 1.13,
"learning_rate": 4.4027176129278815e-05,
"loss": 3.1341,
"step": 2190
},
{
"epoch": 1.13,
"learning_rate": 4.4000972089593566e-05,
"loss": 2.8074,
"step": 2195
},
{
"epoch": 1.13,
"learning_rate": 4.397471852886349e-05,
"loss": 3.0711,
"step": 2200
},
{
"epoch": 1.13,
"learning_rate": 4.3948415515511586e-05,
"loss": 3.0245,
"step": 2205
},
{
"epoch": 1.14,
"learning_rate": 4.3922063118089765e-05,
"loss": 3.0007,
"step": 2210
},
{
"epoch": 1.14,
"learning_rate": 4.389566140527861e-05,
"loss": 2.9818,
"step": 2215
},
{
"epoch": 1.14,
"learning_rate": 4.386921044588727e-05,
"loss": 2.9922,
"step": 2220
},
{
"epoch": 1.14,
"learning_rate": 4.384271030885322e-05,
"loss": 2.8606,
"step": 2225
},
{
"epoch": 1.15,
"learning_rate": 4.381616106324211e-05,
"loss": 2.7615,
"step": 2230
},
{
"epoch": 1.15,
"learning_rate": 4.3789562778247585e-05,
"loss": 2.7838,
"step": 2235
},
{
"epoch": 1.15,
"learning_rate": 4.3762915523191084e-05,
"loss": 2.9275,
"step": 2240
},
{
"epoch": 1.15,
"learning_rate": 4.373621936752168e-05,
"loss": 3.0331,
"step": 2245
},
{
"epoch": 1.16,
"learning_rate": 4.37094743808159e-05,
"loss": 2.9586,
"step": 2250
},
{
"epoch": 1.16,
"learning_rate": 4.3682680632777515e-05,
"loss": 2.7475,
"step": 2255
},
{
"epoch": 1.16,
"learning_rate": 4.36558381932374e-05,
"loss": 3.0034,
"step": 2260
},
{
"epoch": 1.16,
"learning_rate": 4.362894713215334e-05,
"loss": 2.9839,
"step": 2265
},
{
"epoch": 1.17,
"learning_rate": 4.360200751960982e-05,
"loss": 2.8333,
"step": 2270
},
{
"epoch": 1.17,
"learning_rate": 4.3575019425817856e-05,
"loss": 2.853,
"step": 2275
},
{
"epoch": 1.17,
"learning_rate": 4.354798292111483e-05,
"loss": 3.266,
"step": 2280
},
{
"epoch": 1.17,
"learning_rate": 4.35208980759643e-05,
"loss": 3.1553,
"step": 2285
},
{
"epoch": 1.18,
"learning_rate": 4.34937649609558e-05,
"loss": 3.1441,
"step": 2290
},
{
"epoch": 1.18,
"learning_rate": 4.3466583646804674e-05,
"loss": 2.8948,
"step": 2295
},
{
"epoch": 1.18,
"learning_rate": 4.3439354204351866e-05,
"loss": 3.0963,
"step": 2300
},
{
"epoch": 1.18,
"learning_rate": 4.341207670456379e-05,
"loss": 3.0771,
"step": 2305
},
{
"epoch": 1.19,
"learning_rate": 4.3384751218532086e-05,
"loss": 3.0364,
"step": 2310
},
{
"epoch": 1.19,
"learning_rate": 4.335737781747343e-05,
"loss": 2.9928,
"step": 2315
},
{
"epoch": 1.19,
"learning_rate": 4.332995657272944e-05,
"loss": 3.1993,
"step": 2320
},
{
"epoch": 1.19,
"learning_rate": 4.330248755576636e-05,
"loss": 2.9873,
"step": 2325
},
{
"epoch": 1.2,
"learning_rate": 4.3274970838175e-05,
"loss": 2.9258,
"step": 2330
},
{
"epoch": 1.2,
"learning_rate": 4.324740649167044e-05,
"loss": 2.7857,
"step": 2335
},
{
"epoch": 1.2,
"learning_rate": 4.321979458809191e-05,
"loss": 3.1293,
"step": 2340
},
{
"epoch": 1.21,
"learning_rate": 4.31921351994026e-05,
"loss": 3.0588,
"step": 2345
},
{
"epoch": 1.21,
"learning_rate": 4.3164428397689436e-05,
"loss": 3.116,
"step": 2350
},
{
"epoch": 1.21,
"learning_rate": 4.313667425516292e-05,
"loss": 3.0814,
"step": 2355
},
{
"epoch": 1.21,
"learning_rate": 4.310887284415695e-05,
"loss": 2.9781,
"step": 2360
},
{
"epoch": 1.22,
"learning_rate": 4.308102423712859e-05,
"loss": 2.8611,
"step": 2365
},
{
"epoch": 1.22,
"learning_rate": 4.3053128506657934e-05,
"loss": 3.1201,
"step": 2370
},
{
"epoch": 1.22,
"learning_rate": 4.302518572544787e-05,
"loss": 3.108,
"step": 2375
},
{
"epoch": 1.22,
"learning_rate": 4.299719596632392e-05,
"loss": 2.8409,
"step": 2380
},
{
"epoch": 1.23,
"learning_rate": 4.2969159302234044e-05,
"loss": 2.8566,
"step": 2385
},
{
"epoch": 1.23,
"learning_rate": 4.294107580624845e-05,
"loss": 2.7636,
"step": 2390
},
{
"epoch": 1.23,
"learning_rate": 4.291294555155939e-05,
"loss": 3.1108,
"step": 2395
},
{
"epoch": 1.23,
"learning_rate": 4.2884768611481e-05,
"loss": 3.0492,
"step": 2400
},
{
"epoch": 1.24,
"learning_rate": 4.2856545059449063e-05,
"loss": 3.1709,
"step": 2405
},
{
"epoch": 1.24,
"learning_rate": 4.282827496902086e-05,
"loss": 3.0867,
"step": 2410
},
{
"epoch": 1.24,
"learning_rate": 4.279995841387496e-05,
"loss": 2.8054,
"step": 2415
},
{
"epoch": 1.24,
"learning_rate": 4.277159546781103e-05,
"loss": 2.8368,
"step": 2420
},
{
"epoch": 1.25,
"learning_rate": 4.274318620474964e-05,
"loss": 3.0,
"step": 2425
},
{
"epoch": 1.25,
"learning_rate": 4.2714730698732076e-05,
"loss": 2.9218,
"step": 2430
},
{
"epoch": 1.25,
"learning_rate": 4.268622902392014e-05,
"loss": 2.9109,
"step": 2435
},
{
"epoch": 1.25,
"learning_rate": 4.265768125459597e-05,
"loss": 3.074,
"step": 2440
},
{
"epoch": 1.26,
"learning_rate": 4.263480990108211e-05,
"loss": 2.8238,
"step": 2445
},
{
"epoch": 1.26,
"learning_rate": 4.260617934921023e-05,
"loss": 2.9209,
"step": 2450
},
{
"epoch": 1.26,
"learning_rate": 4.257750291145457e-05,
"loss": 2.8323,
"step": 2455
},
{
"epoch": 1.26,
"learning_rate": 4.254878066255274e-05,
"loss": 3.2091,
"step": 2460
},
{
"epoch": 1.27,
"learning_rate": 4.252001267736174e-05,
"loss": 2.8372,
"step": 2465
},
{
"epoch": 1.27,
"learning_rate": 4.2491199030857764e-05,
"loss": 3.0106,
"step": 2470
},
{
"epoch": 1.27,
"learning_rate": 4.246233979813602e-05,
"loss": 2.9969,
"step": 2475
},
{
"epoch": 1.27,
"learning_rate": 4.243343505441052e-05,
"loss": 3.1595,
"step": 2480
},
{
"epoch": 1.28,
"learning_rate": 4.2404484875013895e-05,
"loss": 2.8051,
"step": 2485
},
{
"epoch": 1.28,
"learning_rate": 4.237548933539718e-05,
"loss": 2.8842,
"step": 2490
},
{
"epoch": 1.28,
"learning_rate": 4.234644851112964e-05,
"loss": 2.9729,
"step": 2495
},
{
"epoch": 1.28,
"learning_rate": 4.231736247789856e-05,
"loss": 2.7805,
"step": 2500
},
{
"epoch": 1.29,
"learning_rate": 4.228823131150904e-05,
"loss": 2.9725,
"step": 2505
},
{
"epoch": 1.29,
"learning_rate": 4.225905508788382e-05,
"loss": 3.0599,
"step": 2510
},
{
"epoch": 1.29,
"learning_rate": 4.222983388306308e-05,
"loss": 2.7567,
"step": 2515
},
{
"epoch": 1.29,
"learning_rate": 4.2200567773204194e-05,
"loss": 2.884,
"step": 2520
},
{
"epoch": 1.3,
"learning_rate": 4.217125683458161e-05,
"loss": 2.8594,
"step": 2525
},
{
"epoch": 1.3,
"learning_rate": 4.214190114358659e-05,
"loss": 2.9515,
"step": 2530
},
{
"epoch": 1.3,
"learning_rate": 4.211250077672704e-05,
"loss": 2.992,
"step": 2535
},
{
"epoch": 1.31,
"learning_rate": 4.208305581062728e-05,
"loss": 3.1784,
"step": 2540
},
{
"epoch": 1.31,
"learning_rate": 4.205356632202789e-05,
"loss": 3.0196,
"step": 2545
},
{
"epoch": 1.31,
"learning_rate": 4.20240323877855e-05,
"loss": 2.7891,
"step": 2550
},
{
"epoch": 1.31,
"learning_rate": 4.1994454084872525e-05,
"loss": 2.8974,
"step": 2555
},
{
"epoch": 1.32,
"learning_rate": 4.196483149037707e-05,
"loss": 2.8219,
"step": 2560
},
{
"epoch": 1.32,
"learning_rate": 4.1935164681502624e-05,
"loss": 2.8833,
"step": 2565
},
{
"epoch": 1.32,
"learning_rate": 4.190545373556796e-05,
"loss": 3.0166,
"step": 2570
},
{
"epoch": 1.32,
"learning_rate": 4.187569873000684e-05,
"loss": 3.1676,
"step": 2575
},
{
"epoch": 1.33,
"learning_rate": 4.184589974236788e-05,
"loss": 2.9239,
"step": 2580
},
{
"epoch": 1.33,
"learning_rate": 4.181605685031432e-05,
"loss": 3.0094,
"step": 2585
},
{
"epoch": 1.33,
"learning_rate": 4.1786170131623825e-05,
"loss": 3.0235,
"step": 2590
},
{
"epoch": 1.33,
"learning_rate": 4.1756239664188273e-05,
"loss": 3.058,
"step": 2595
},
{
"epoch": 1.34,
"learning_rate": 4.172626552601358e-05,
"loss": 2.7391,
"step": 2600
},
{
"epoch": 1.34,
"learning_rate": 4.169624779521944e-05,
"loss": 2.9975,
"step": 2605
},
{
"epoch": 1.34,
"learning_rate": 4.1666186550039214e-05,
"loss": 2.7703,
"step": 2610
},
{
"epoch": 1.34,
"learning_rate": 4.163608186881964e-05,
"loss": 2.8288,
"step": 2615
},
{
"epoch": 1.35,
"learning_rate": 4.160593383002066e-05,
"loss": 2.9945,
"step": 2620
},
{
"epoch": 1.35,
"learning_rate": 4.157574251221521e-05,
"loss": 2.8768,
"step": 2625
},
{
"epoch": 1.35,
"learning_rate": 4.1545507994089064e-05,
"loss": 2.7161,
"step": 2630
},
{
"epoch": 1.35,
"learning_rate": 4.1515230354440524e-05,
"loss": 3.0821,
"step": 2635
},
{
"epoch": 1.36,
"learning_rate": 4.148490967218033e-05,
"loss": 3.1119,
"step": 2640
},
{
"epoch": 1.36,
"learning_rate": 4.145454602633137e-05,
"loss": 2.8815,
"step": 2645
},
{
"epoch": 1.36,
"learning_rate": 4.14241394960285e-05,
"loss": 2.9891,
"step": 2650
},
{
"epoch": 1.36,
"learning_rate": 4.1393690160518384e-05,
"loss": 3.2624,
"step": 2655
},
{
"epoch": 1.37,
"learning_rate": 4.1363198099159205e-05,
"loss": 2.7588,
"step": 2660
},
{
"epoch": 1.37,
"learning_rate": 4.1332663391420515e-05,
"loss": 3.0405,
"step": 2665
},
{
"epoch": 1.37,
"learning_rate": 4.130208611688302e-05,
"loss": 2.9254,
"step": 2670
},
{
"epoch": 1.37,
"learning_rate": 4.1271466355238354e-05,
"loss": 2.8499,
"step": 2675
},
{
"epoch": 1.38,
"learning_rate": 4.124080418628888e-05,
"loss": 3.0828,
"step": 2680
},
{
"epoch": 1.38,
"learning_rate": 4.1210099689947513e-05,
"loss": 2.8524,
"step": 2685
},
{
"epoch": 1.38,
"learning_rate": 4.117935294623743e-05,
"loss": 3.0711,
"step": 2690
},
{
"epoch": 1.38,
"learning_rate": 4.1148564035291956e-05,
"loss": 3.0343,
"step": 2695
},
{
"epoch": 1.39,
"learning_rate": 4.1117733037354314e-05,
"loss": 2.8869,
"step": 2700
},
{
"epoch": 1.39,
"learning_rate": 4.1086860032777395e-05,
"loss": 3.0381,
"step": 2705
},
{
"epoch": 1.39,
"learning_rate": 4.105594510202359e-05,
"loss": 2.9049,
"step": 2710
},
{
"epoch": 1.4,
"learning_rate": 4.102498832566454e-05,
"loss": 3.0071,
"step": 2715
},
{
"epoch": 1.4,
"learning_rate": 4.0993989784380947e-05,
"loss": 2.9955,
"step": 2720
},
{
"epoch": 1.4,
"learning_rate": 4.096294955896239e-05,
"loss": 2.9568,
"step": 2725
},
{
"epoch": 1.4,
"learning_rate": 4.093186773030705e-05,
"loss": 2.9727,
"step": 2730
},
{
"epoch": 1.41,
"learning_rate": 4.090074437942155e-05,
"loss": 3.0501,
"step": 2735
},
{
"epoch": 1.41,
"learning_rate": 4.086957958742075e-05,
"loss": 3.2328,
"step": 2740
},
{
"epoch": 1.41,
"learning_rate": 4.0838373435527494e-05,
"loss": 2.7966,
"step": 2745
},
{
"epoch": 1.41,
"learning_rate": 4.08071260050724e-05,
"loss": 2.8666,
"step": 2750
},
{
"epoch": 1.42,
"learning_rate": 4.077583737749373e-05,
"loss": 2.9765,
"step": 2755
},
{
"epoch": 1.42,
"learning_rate": 4.074450763433704e-05,
"loss": 2.6923,
"step": 2760
},
{
"epoch": 1.42,
"learning_rate": 4.07131368572551e-05,
"loss": 2.9031,
"step": 2765
},
{
"epoch": 1.42,
"learning_rate": 4.06817251280076e-05,
"loss": 2.8287,
"step": 2770
},
{
"epoch": 1.43,
"learning_rate": 4.065027252846095e-05,
"loss": 3.1046,
"step": 2775
},
{
"epoch": 1.43,
"learning_rate": 4.0618779140588124e-05,
"loss": 3.0414,
"step": 2780
},
{
"epoch": 1.43,
"learning_rate": 4.058724504646834e-05,
"loss": 2.8644,
"step": 2785
},
{
"epoch": 1.43,
"learning_rate": 4.055567032828695e-05,
"loss": 2.9589,
"step": 2790
},
{
"epoch": 1.44,
"learning_rate": 4.052405506833516e-05,
"loss": 2.886,
"step": 2795
},
{
"epoch": 1.44,
"learning_rate": 4.0492399349009846e-05,
"loss": 2.8026,
"step": 2800
},
{
"epoch": 1.44,
"learning_rate": 4.0460703252813326e-05,
"loss": 3.0376,
"step": 2805
},
{
"epoch": 1.44,
"learning_rate": 4.042896686235316e-05,
"loss": 2.9751,
"step": 2810
},
{
"epoch": 1.45,
"learning_rate": 4.039719026034191e-05,
"loss": 2.8638,
"step": 2815
},
{
"epoch": 1.45,
"learning_rate": 4.036537352959694e-05,
"loss": 3.0403,
"step": 2820
},
{
"epoch": 1.45,
"learning_rate": 4.033351675304022e-05,
"loss": 2.8241,
"step": 2825
},
{
"epoch": 1.45,
"learning_rate": 4.030162001369807e-05,
"loss": 3.1196,
"step": 2830
},
{
"epoch": 1.46,
"learning_rate": 4.026968339470096e-05,
"loss": 2.8388,
"step": 2835
},
{
"epoch": 1.46,
"learning_rate": 4.0237706979283304e-05,
"loss": 2.9974,
"step": 2840
},
{
"epoch": 1.46,
"learning_rate": 4.020569085078324e-05,
"loss": 3.0178,
"step": 2845
},
{
"epoch": 1.46,
"learning_rate": 4.0173635092642386e-05,
"loss": 3.0166,
"step": 2850
},
{
"epoch": 1.47,
"learning_rate": 4.014153978840567e-05,
"loss": 3.0657,
"step": 2855
},
{
"epoch": 1.47,
"learning_rate": 4.010940502172108e-05,
"loss": 3.0229,
"step": 2860
},
{
"epoch": 1.47,
"learning_rate": 4.007723087633943e-05,
"loss": 2.5904,
"step": 2865
},
{
"epoch": 1.47,
"learning_rate": 4.0045017436114205e-05,
"loss": 2.8607,
"step": 2870
},
{
"epoch": 1.48,
"learning_rate": 4.001276478500127e-05,
"loss": 3.0797,
"step": 2875
},
{
"epoch": 1.48,
"learning_rate": 3.998047300705868e-05,
"loss": 2.9475,
"step": 2880
},
{
"epoch": 1.48,
"learning_rate": 3.9948142186446486e-05,
"loss": 2.7829,
"step": 2885
},
{
"epoch": 1.49,
"learning_rate": 3.991577240742649e-05,
"loss": 3.0715,
"step": 2890
},
{
"epoch": 1.49,
"learning_rate": 3.9883363754362004e-05,
"loss": 3.0126,
"step": 2895
},
{
"epoch": 1.49,
"learning_rate": 3.985091631171769e-05,
"loss": 3.0835,
"step": 2900
},
{
"epoch": 1.49,
"learning_rate": 3.981843016405927e-05,
"loss": 2.9109,
"step": 2905
},
{
"epoch": 1.5,
"learning_rate": 3.978590539605338e-05,
"loss": 3.0058,
"step": 2910
},
{
"epoch": 1.5,
"learning_rate": 3.975334209246727e-05,
"loss": 2.7376,
"step": 2915
},
{
"epoch": 1.5,
"learning_rate": 3.972074033816866e-05,
"loss": 2.7173,
"step": 2920
},
{
"epoch": 1.5,
"learning_rate": 3.968810021812544e-05,
"loss": 2.8642,
"step": 2925
},
{
"epoch": 1.51,
"learning_rate": 3.965542181740553e-05,
"loss": 2.8643,
"step": 2930
},
{
"epoch": 1.51,
"learning_rate": 3.9622705221176596e-05,
"loss": 3.1752,
"step": 2935
},
{
"epoch": 1.51,
"learning_rate": 3.958995051470585e-05,
"loss": 2.7045,
"step": 2940
},
{
"epoch": 1.51,
"learning_rate": 3.955715778335984e-05,
"loss": 3.0828,
"step": 2945
},
{
"epoch": 1.52,
"learning_rate": 3.952432711260421e-05,
"loss": 2.982,
"step": 2950
},
{
"epoch": 1.52,
"learning_rate": 3.949145858800348e-05,
"loss": 2.9762,
"step": 2955
},
{
"epoch": 1.52,
"learning_rate": 3.945855229522084e-05,
"loss": 3.1026,
"step": 2960
},
{
"epoch": 1.52,
"learning_rate": 3.942560832001789e-05,
"loss": 3.1225,
"step": 2965
},
{
"epoch": 1.53,
"learning_rate": 3.9392626748254456e-05,
"loss": 3.0462,
"step": 2970
},
{
"epoch": 1.53,
"learning_rate": 3.9359607665888346e-05,
"loss": 2.8954,
"step": 2975
},
{
"epoch": 1.53,
"learning_rate": 3.932655115897513e-05,
"loss": 2.8361,
"step": 2980
},
{
"epoch": 1.53,
"learning_rate": 3.92934573136679e-05,
"loss": 3.1644,
"step": 2985
},
{
"epoch": 1.54,
"learning_rate": 3.9260326216217105e-05,
"loss": 3.0226,
"step": 2990
},
{
"epoch": 1.54,
"learning_rate": 3.922715795297022e-05,
"loss": 3.0285,
"step": 2995
},
{
"epoch": 1.54,
"learning_rate": 3.919395261037163e-05,
"loss": 3.0582,
"step": 3000
},
{
"epoch": 1.54,
"learning_rate": 3.9160710274962345e-05,
"loss": 2.9447,
"step": 3005
},
{
"epoch": 1.55,
"learning_rate": 3.912743103337977e-05,
"loss": 3.0433,
"step": 3010
},
{
"epoch": 1.55,
"learning_rate": 3.9094114972357524e-05,
"loss": 3.0276,
"step": 3015
},
{
"epoch": 1.55,
"learning_rate": 3.9060762178725165e-05,
"loss": 3.1617,
"step": 3020
},
{
"epoch": 1.55,
"learning_rate": 3.9027372739407994e-05,
"loss": 2.962,
"step": 3025
},
{
"epoch": 1.56,
"learning_rate": 3.899394674142682e-05,
"loss": 2.9722,
"step": 3030
},
{
"epoch": 1.56,
"learning_rate": 3.8960484271897736e-05,
"loss": 2.9605,
"step": 3035
},
{
"epoch": 1.56,
"learning_rate": 3.892698541803187e-05,
"loss": 3.1036,
"step": 3040
},
{
"epoch": 1.56,
"learning_rate": 3.889345026713519e-05,
"loss": 3.1859,
"step": 3045
},
{
"epoch": 1.57,
"learning_rate": 3.8859878906608275e-05,
"loss": 3.0349,
"step": 3050
},
{
"epoch": 1.57,
"learning_rate": 3.882627142394605e-05,
"loss": 2.9102,
"step": 3055
},
{
"epoch": 1.57,
"learning_rate": 3.87926279067376e-05,
"loss": 2.9039,
"step": 3060
},
{
"epoch": 1.58,
"learning_rate": 3.8758948442665914e-05,
"loss": 3.0984,
"step": 3065
},
{
"epoch": 1.58,
"learning_rate": 3.872523311950768e-05,
"loss": 2.9782,
"step": 3070
},
{
"epoch": 1.58,
"learning_rate": 3.869148202513303e-05,
"loss": 2.7941,
"step": 3075
},
{
"epoch": 1.58,
"learning_rate": 3.8657695247505325e-05,
"loss": 3.1176,
"step": 3080
},
{
"epoch": 1.59,
"learning_rate": 3.862387287468094e-05,
"loss": 2.8274,
"step": 3085
},
{
"epoch": 1.59,
"learning_rate": 3.859001499480901e-05,
"loss": 3.0035,
"step": 3090
},
{
"epoch": 1.59,
"learning_rate": 3.8556121696131194e-05,
"loss": 3.0127,
"step": 3095
},
{
"epoch": 1.59,
"learning_rate": 3.852219306698149e-05,
"loss": 3.0512,
"step": 3100
},
{
"epoch": 1.6,
"learning_rate": 3.848822919578596e-05,
"loss": 2.8802,
"step": 3105
},
{
"epoch": 1.6,
"learning_rate": 3.845423017106251e-05,
"loss": 2.9527,
"step": 3110
},
{
"epoch": 1.6,
"learning_rate": 3.842019608142069e-05,
"loss": 2.9869,
"step": 3115
},
{
"epoch": 1.6,
"learning_rate": 3.838612701556138e-05,
"loss": 2.7298,
"step": 3120
},
{
"epoch": 1.61,
"learning_rate": 3.835202306227669e-05,
"loss": 3.0647,
"step": 3125
},
{
"epoch": 1.61,
"learning_rate": 3.8317884310449614e-05,
"loss": 2.9665,
"step": 3130
},
{
"epoch": 1.61,
"learning_rate": 3.8283710849053834e-05,
"loss": 2.9411,
"step": 3135
},
{
"epoch": 1.61,
"learning_rate": 3.824950276715351e-05,
"loss": 2.8853,
"step": 3140
},
{
"epoch": 1.62,
"learning_rate": 3.8215260153903024e-05,
"loss": 2.7117,
"step": 3145
},
{
"epoch": 1.62,
"learning_rate": 3.818098309854676e-05,
"loss": 3.1676,
"step": 3150
},
{
"epoch": 1.62,
"learning_rate": 3.8146671690418865e-05,
"loss": 2.9812,
"step": 3155
},
{
"epoch": 1.62,
"learning_rate": 3.811232601894301e-05,
"loss": 2.8832,
"step": 3160
},
{
"epoch": 1.63,
"learning_rate": 3.8077946173632175e-05,
"loss": 3.0479,
"step": 3165
},
{
"epoch": 1.63,
"learning_rate": 3.80435322440884e-05,
"loss": 2.9631,
"step": 3170
},
{
"epoch": 1.63,
"learning_rate": 3.800908432000254e-05,
"loss": 2.9479,
"step": 3175
},
{
"epoch": 1.63,
"learning_rate": 3.7974602491154094e-05,
"loss": 2.752,
"step": 3180
},
{
"epoch": 1.64,
"learning_rate": 3.7940086847410874e-05,
"loss": 3.0117,
"step": 3185
},
{
"epoch": 1.64,
"learning_rate": 3.7905537478728845e-05,
"loss": 3.0633,
"step": 3190
},
{
"epoch": 1.64,
"learning_rate": 3.787095447515187e-05,
"loss": 3.1068,
"step": 3195
},
{
"epoch": 1.64,
"learning_rate": 3.783633792681147e-05,
"loss": 2.9438,
"step": 3200
},
{
"epoch": 1.65,
"learning_rate": 3.7801687923926577e-05,
"loss": 2.9187,
"step": 3205
},
{
"epoch": 1.65,
"learning_rate": 3.7767004556803346e-05,
"loss": 3.0672,
"step": 3210
},
{
"epoch": 1.65,
"learning_rate": 3.773228791583484e-05,
"loss": 2.9911,
"step": 3215
},
{
"epoch": 1.65,
"learning_rate": 3.76975380915009e-05,
"loss": 3.0206,
"step": 3220
},
{
"epoch": 1.66,
"learning_rate": 3.7662755174367795e-05,
"loss": 3.1758,
"step": 3225
},
{
"epoch": 1.66,
"learning_rate": 3.7627939255088064e-05,
"loss": 2.5882,
"step": 3230
},
{
"epoch": 1.66,
"learning_rate": 3.759309042440028e-05,
"loss": 3.1835,
"step": 3235
},
{
"epoch": 1.66,
"learning_rate": 3.755820877312876e-05,
"loss": 3.0573,
"step": 3240
},
{
"epoch": 1.67,
"learning_rate": 3.752329439218337e-05,
"loss": 2.9557,
"step": 3245
},
{
"epoch": 1.67,
"learning_rate": 3.748834737255927e-05,
"loss": 2.7683,
"step": 3250
},
{
"epoch": 1.67,
"learning_rate": 3.74533678053367e-05,
"loss": 2.9666,
"step": 3255
},
{
"epoch": 1.68,
"learning_rate": 3.741835578168071e-05,
"loss": 3.1286,
"step": 3260
},
{
"epoch": 1.68,
"learning_rate": 3.738331139284096e-05,
"loss": 2.8983,
"step": 3265
},
{
"epoch": 1.68,
"learning_rate": 3.734823473015143e-05,
"loss": 2.9258,
"step": 3270
},
{
"epoch": 1.68,
"learning_rate": 3.7313125885030244e-05,
"loss": 3.2793,
"step": 3275
},
{
"epoch": 1.69,
"learning_rate": 3.727798494897938e-05,
"loss": 2.7685,
"step": 3280
},
{
"epoch": 1.69,
"learning_rate": 3.724281201358447e-05,
"loss": 2.7007,
"step": 3285
},
{
"epoch": 1.69,
"learning_rate": 3.7207607170514534e-05,
"loss": 3.0827,
"step": 3290
},
{
"epoch": 1.69,
"learning_rate": 3.717237051152175e-05,
"loss": 3.1145,
"step": 3295
},
{
"epoch": 1.7,
"learning_rate": 3.7137102128441216e-05,
"loss": 2.9999,
"step": 3300
}
],
"logging_steps": 5,
"max_steps": 9730,
"num_input_tokens_seen": 0,
"num_train_epochs": 5,
"save_steps": 300,
"total_flos": 3.895031732436173e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}