916 lines
18 KiB
JSON
916 lines
18 KiB
JSON
{
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 2.3047177107501935,
|
|
"eval_steps": 500,
|
|
"global_step": 1490,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 0.000199739323151795,
|
|
"loss": 1.5905,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 0.00019682229406025635,
|
|
"loss": 1.3988,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 0.00019075754196709572,
|
|
"loss": 1.4513,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 0.00018174223385588917,
|
|
"loss": 1.3878,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 0.00016949152542372882,
|
|
"loss": 1.2662,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 0.0001999998602293167,
|
|
"loss": 1.0731,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 0.0001999830882200816,
|
|
"loss": 1.0733,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 0.0001999383674462943,
|
|
"loss": 1.0432,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 0.00019986571040897272,
|
|
"loss": 1.0459,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 0.00019976513741829603,
|
|
"loss": 1.0564,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 0.00019963667658792704,
|
|
"loss": 1.006,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 0.00019948036382715371,
|
|
"loss": 1.0124,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 0.0001992962428308511,
|
|
"loss": 1.0495,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 0.00019908436506726714,
|
|
"loss": 1.0061,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 0.00019884478976363548,
|
|
"loss": 1.0166,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 0.00019857758388961943,
|
|
"loss": 0.986,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 0.0001982828221385916,
|
|
"loss": 1.0052,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 0.00019796058690675435,
|
|
"loss": 0.9986,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 0.0001976109682701075,
|
|
"loss": 0.989,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 0.00019723406395926856,
|
|
"loss": 1.0184,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 0.00019682997933215385,
|
|
"loss": 0.9896,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 0.00019639882734452722,
|
|
"loss": 1.0046,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 0.000195940728518425,
|
|
"loss": 0.9922,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 0.00019545581090846584,
|
|
"loss": 0.9665,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 0.00019494421006605492,
|
|
"loss": 1.0012,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 0.00019440606900149275,
|
|
"loss": 0.9937,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 0.0001938415381439987,
|
|
"loss": 0.984,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 0.00019325077529966077,
|
|
"loss": 0.9651,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 0.00019263394560732326,
|
|
"loss": 0.9621,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 0.00019199122149242485,
|
|
"loss": 0.9761,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 0.00019132278261879945,
|
|
"loss": 0.9792,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 0.00019062881583845398,
|
|
"loss": 0.9918,
|
|
"step": 320
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 0.00018990951513933683,
|
|
"loss": 0.969,
|
|
"step": 330
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 0.00018916508159111126,
|
|
"loss": 0.9513,
|
|
"step": 340
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 0.0001883957232889495,
|
|
"loss": 0.9987,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 0.00018760165529536286,
|
|
"loss": 0.9687,
|
|
"step": 360
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 0.00018678309958008432,
|
|
"loss": 0.9734,
|
|
"step": 370
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 0.00018594028495802015,
|
|
"loss": 0.9705,
|
|
"step": 380
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 0.00018507344702528816,
|
|
"loss": 0.949,
|
|
"step": 390
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 0.0001841828280933604,
|
|
"loss": 0.9709,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 0.00018326867712132865,
|
|
"loss": 0.9484,
|
|
"step": 410
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 0.00018233124964631156,
|
|
"loss": 0.9677,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 0.00018137080771202325,
|
|
"loss": 0.9597,
|
|
"step": 430
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 0.00018038761979552285,
|
|
"loss": 0.9289,
|
|
"step": 440
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 0.00017938196073216586,
|
|
"loss": 0.9749,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 0.00017835411163877782,
|
|
"loss": 0.9445,
|
|
"step": 460
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 0.00017730435983507277,
|
|
"loss": 0.9609,
|
|
"step": 470
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 0.00017623299876333697,
|
|
"loss": 0.9652,
|
|
"step": 480
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 0.0001751403279064018,
|
|
"loss": 0.9415,
|
|
"step": 490
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 0.0001740266527039276,
|
|
"loss": 0.9769,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 0.00017289228446702292,
|
|
"loss": 0.9477,
|
|
"step": 510
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 0.00017173754029122188,
|
|
"loss": 0.9622,
|
|
"step": 520
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 0.00017056274296784518,
|
|
"loss": 0.9631,
|
|
"step": 530
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 0.00016936822089376837,
|
|
"loss": 0.9482,
|
|
"step": 540
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 0.00016815430797962374,
|
|
"loss": 0.9666,
|
|
"step": 550
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 0.00016692134355646046,
|
|
"loss": 0.9331,
|
|
"step": 560
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 0.00016566967228088972,
|
|
"loss": 0.9395,
|
|
"step": 570
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 0.0001643996440387412,
|
|
"loss": 0.9432,
|
|
"step": 580
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 0.0001631116138472578,
|
|
"loss": 0.9244,
|
|
"step": 590
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 0.00016180594175585586,
|
|
"loss": 0.9617,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 0.00016048299274547883,
|
|
"loss": 0.9213,
|
|
"step": 610
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 0.00015914313662657224,
|
|
"loss": 0.9497,
|
|
"step": 620
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 0.00015778674793570896,
|
|
"loss": 0.9377,
|
|
"step": 630
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 0.00015641420583089295,
|
|
"loss": 0.9205,
|
|
"step": 640
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 0.00015502589398557146,
|
|
"loss": 0.9664,
|
|
"step": 650
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 0.0001536222004813849,
|
|
"loss": 0.86,
|
|
"step": 660
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 0.0001522035176996845,
|
|
"loss": 0.8729,
|
|
"step": 670
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 0.00015077024221184793,
|
|
"loss": 0.8386,
|
|
"step": 680
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 0.00014932277466842377,
|
|
"loss": 0.836,
|
|
"step": 690
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 0.0001478615196871358,
|
|
"loss": 0.8874,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 0.00014638688573977805,
|
|
"loss": 0.8546,
|
|
"step": 710
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 0.00014489928503803285,
|
|
"loss": 0.8682,
|
|
"step": 720
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 0.00014339913341824314,
|
|
"loss": 0.8407,
|
|
"step": 730
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 0.0001418868502251717,
|
|
"loss": 0.8476,
|
|
"step": 740
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 0.00014036285819477967,
|
|
"loss": 0.8962,
|
|
"step": 750
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 0.0001388275833360572,
|
|
"loss": 0.8799,
|
|
"step": 760
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 0.00013728145481193882,
|
|
"loss": 0.8585,
|
|
"step": 770
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 0.00013572490481933758,
|
|
"loss": 0.8541,
|
|
"step": 780
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 0.000134158368468331,
|
|
"loss": 0.8239,
|
|
"step": 790
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 0.0001325822836605323,
|
|
"loss": 0.8758,
|
|
"step": 800
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 0.00013099709096668193,
|
|
"loss": 0.8682,
|
|
"step": 810
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 0.00012940323350349258,
|
|
"loss": 0.8588,
|
|
"step": 820
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 0.0001278011568097824,
|
|
"loss": 0.8455,
|
|
"step": 830
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 0.00012619130872193162,
|
|
"loss": 0.8485,
|
|
"step": 840
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 0.00012457413924869644,
|
|
"loss": 0.8995,
|
|
"step": 850
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 0.0001229501004454159,
|
|
"loss": 0.8736,
|
|
"step": 860
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 0.00012131964628764678,
|
|
"loss": 0.8638,
|
|
"step": 870
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 0.00011968323254426135,
|
|
"loss": 0.8412,
|
|
"step": 880
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 0.00011804131665004423,
|
|
"loss": 0.815,
|
|
"step": 890
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 0.00011639435757782336,
|
|
"loss": 0.8969,
|
|
"step": 900
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 0.0001147428157101709,
|
|
"loss": 0.8737,
|
|
"step": 910
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 0.00011308715271071049,
|
|
"loss": 0.8604,
|
|
"step": 920
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 0.00011142783139506601,
|
|
"loss": 0.8425,
|
|
"step": 930
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 0.00010976531560148841,
|
|
"loss": 0.8262,
|
|
"step": 940
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 0.00010810007006119685,
|
|
"loss": 0.9044,
|
|
"step": 950
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 0.00010643256026846992,
|
|
"loss": 0.8575,
|
|
"step": 960
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 0.00010476325235052389,
|
|
"loss": 0.8602,
|
|
"step": 970
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 0.00010309261293721384,
|
|
"loss": 0.8507,
|
|
"step": 980
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 0.00010142110903059424,
|
|
"loss": 0.8279,
|
|
"step": 990
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 9.974920787437567e-05,
|
|
"loss": 0.8845,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 9.807737682331383e-05,
|
|
"loss": 0.8515,
|
|
"step": 1010
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 9.640608321256761e-05,
|
|
"loss": 0.8467,
|
|
"step": 1020
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 9.47357942270625e-05,
|
|
"loss": 0.8406,
|
|
"step": 1030
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 9.306697677089621e-05,
|
|
"loss": 0.8435,
|
|
"step": 1040
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 9.140009733682262e-05,
|
|
"loss": 0.8836,
|
|
"step": 1050
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 8.973562187585071e-05,
|
|
"loss": 0.866,
|
|
"step": 1060
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 8.80740156669951e-05,
|
|
"loss": 0.8585,
|
|
"step": 1070
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 8.641574318721413e-05,
|
|
"loss": 0.847,
|
|
"step": 1080
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 8.476126798157258e-05,
|
|
"loss": 0.8221,
|
|
"step": 1090
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 8.311105253366448e-05,
|
|
"loss": 0.8873,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 8.146555813633274e-05,
|
|
"loss": 0.8415,
|
|
"step": 1110
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 7.982524476272188e-05,
|
|
"loss": 0.8533,
|
|
"step": 1120
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 7.819057093769931e-05,
|
|
"loss": 0.8281,
|
|
"step": 1130
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 7.656199360968167e-05,
|
|
"loss": 0.8319,
|
|
"step": 1140
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 7.493996802290186e-05,
|
|
"loss": 0.8593,
|
|
"step": 1150
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 7.332494759015226e-05,
|
|
"loss": 0.8361,
|
|
"step": 1160
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 7.171738376604012e-05,
|
|
"loss": 0.8536,
|
|
"step": 1170
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 7.011772592079013e-05,
|
|
"loss": 0.8478,
|
|
"step": 1180
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 6.85264212146299e-05,
|
|
"loss": 0.8276,
|
|
"step": 1190
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 6.694391447279287e-05,
|
|
"loss": 0.8934,
|
|
"step": 1200
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 6.537064806117435e-05,
|
|
"loss": 0.8359,
|
|
"step": 1210
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 6.380706176267455e-05,
|
|
"loss": 0.8477,
|
|
"step": 1220
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 6.225359265426414e-05,
|
|
"loss": 0.8293,
|
|
"step": 1230
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 6.071067498480583e-05,
|
|
"loss": 0.828,
|
|
"step": 1240
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5.9178740053666706e-05,
|
|
"loss": 0.8773,
|
|
"step": 1250
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5.765821609015495e-05,
|
|
"loss": 0.8376,
|
|
"step": 1260
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5.61495281338148e-05,
|
|
"loss": 0.8384,
|
|
"step": 1270
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5.465309791561328e-05,
|
|
"loss": 0.8403,
|
|
"step": 1280
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"learning_rate": 5.316934374005142e-05,
|
|
"loss": 0.8464,
|
|
"step": 1290
|
|
},
|
|
{
|
|
"epoch": 2.01,
|
|
"learning_rate": 5.1698680368233864e-05,
|
|
"loss": 0.8334,
|
|
"step": 1300
|
|
},
|
|
{
|
|
"epoch": 2.03,
|
|
"learning_rate": 5.02415189019283e-05,
|
|
"loss": 0.7771,
|
|
"step": 1310
|
|
},
|
|
{
|
|
"epoch": 2.04,
|
|
"learning_rate": 4.879826666864852e-05,
|
|
"loss": 0.7607,
|
|
"step": 1320
|
|
},
|
|
{
|
|
"epoch": 2.06,
|
|
"learning_rate": 4.736932710779173e-05,
|
|
"loss": 0.7367,
|
|
"step": 1330
|
|
},
|
|
{
|
|
"epoch": 2.07,
|
|
"learning_rate": 4.595509965786368e-05,
|
|
"loss": 0.763,
|
|
"step": 1340
|
|
},
|
|
{
|
|
"epoch": 2.09,
|
|
"learning_rate": 4.4555979644821145e-05,
|
|
"loss": 0.7878,
|
|
"step": 1350
|
|
},
|
|
{
|
|
"epoch": 2.1,
|
|
"learning_rate": 4.3172358171564916e-05,
|
|
"loss": 0.763,
|
|
"step": 1360
|
|
},
|
|
{
|
|
"epoch": 2.12,
|
|
"learning_rate": 4.180462200861256e-05,
|
|
"loss": 0.7487,
|
|
"step": 1370
|
|
},
|
|
{
|
|
"epoch": 2.13,
|
|
"learning_rate": 4.045315348598283e-05,
|
|
"loss": 0.738,
|
|
"step": 1380
|
|
},
|
|
{
|
|
"epoch": 2.15,
|
|
"learning_rate": 3.9118330386320814e-05,
|
|
"loss": 0.7439,
|
|
"step": 1390
|
|
},
|
|
{
|
|
"epoch": 2.17,
|
|
"learning_rate": 3.780052583929475e-05,
|
|
"loss": 0.7903,
|
|
"step": 1400
|
|
},
|
|
{
|
|
"epoch": 2.18,
|
|
"learning_rate": 3.650010821729303e-05,
|
|
"loss": 0.7653,
|
|
"step": 1410
|
|
},
|
|
{
|
|
"epoch": 2.2,
|
|
"learning_rate": 3.521744103245145e-05,
|
|
"loss": 0.7447,
|
|
"step": 1420
|
|
},
|
|
{
|
|
"epoch": 2.21,
|
|
"learning_rate": 3.395288283503867e-05,
|
|
"loss": 0.7186,
|
|
"step": 1430
|
|
},
|
|
{
|
|
"epoch": 2.23,
|
|
"learning_rate": 3.2706787113229075e-05,
|
|
"loss": 0.7259,
|
|
"step": 1440
|
|
},
|
|
{
|
|
"epoch": 2.24,
|
|
"learning_rate": 3.147950219429036e-05,
|
|
"loss": 0.8041,
|
|
"step": 1450
|
|
},
|
|
{
|
|
"epoch": 2.26,
|
|
"learning_rate": 3.027137114721401e-05,
|
|
"loss": 0.7545,
|
|
"step": 1460
|
|
},
|
|
{
|
|
"epoch": 2.27,
|
|
"learning_rate": 2.9082731686815412e-05,
|
|
"loss": 0.7363,
|
|
"step": 1470
|
|
},
|
|
{
|
|
"epoch": 2.29,
|
|
"learning_rate": 2.7913916079330905e-05,
|
|
"loss": 0.7364,
|
|
"step": 1480
|
|
},
|
|
{
|
|
"epoch": 2.3,
|
|
"learning_rate": 2.676525104953759e-05,
|
|
"loss": 0.7509,
|
|
"step": 1490
|
|
}
|
|
],
|
|
"logging_steps": 10,
|
|
"max_steps": 1938,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 3,
|
|
"save_steps": 10,
|
|
"total_flos": 4.7276257552131686e+17,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|