dumb-dev's picture
at least i generates some kind of output now...
8bb1997 verified
raw
history blame contribute delete
No virus
47.5 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 5.0,
"eval_steps": 500,
"global_step": 3710,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.9999377559783764e-05,
"loss": 1.7762,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 4.999751027012961e-05,
"loss": 1.7656,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 4.999439822401963e-05,
"loss": 1.8093,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 4.9990041576418836e-05,
"loss": 1.5602,
"step": 40
},
{
"epoch": 0.07,
"learning_rate": 4.998444054426743e-05,
"loss": 1.5704,
"step": 50
},
{
"epoch": 0.08,
"learning_rate": 4.9977595406470044e-05,
"loss": 1.7002,
"step": 60
},
{
"epoch": 0.09,
"learning_rate": 4.9969506503881794e-05,
"loss": 1.694,
"step": 70
},
{
"epoch": 0.11,
"learning_rate": 4.996017423929133e-05,
"loss": 1.6043,
"step": 80
},
{
"epoch": 0.12,
"learning_rate": 4.9949599077400825e-05,
"loss": 1.6391,
"step": 90
},
{
"epoch": 0.13,
"learning_rate": 4.9937781544802745e-05,
"loss": 1.6518,
"step": 100
},
{
"epoch": 0.15,
"learning_rate": 4.992472222995368e-05,
"loss": 1.6928,
"step": 110
},
{
"epoch": 0.16,
"learning_rate": 4.9910421783145074e-05,
"loss": 1.7065,
"step": 120
},
{
"epoch": 0.18,
"learning_rate": 4.9894880916470764e-05,
"loss": 1.6429,
"step": 130
},
{
"epoch": 0.19,
"learning_rate": 4.98781004037916e-05,
"loss": 1.5974,
"step": 140
},
{
"epoch": 0.2,
"learning_rate": 4.986008108069685e-05,
"loss": 1.6029,
"step": 150
},
{
"epoch": 0.22,
"learning_rate": 4.984082384446263e-05,
"loss": 1.4892,
"step": 160
},
{
"epoch": 0.23,
"learning_rate": 4.982032965400719e-05,
"loss": 1.4642,
"step": 170
},
{
"epoch": 0.24,
"learning_rate": 4.97985995298432e-05,
"loss": 1.5265,
"step": 180
},
{
"epoch": 0.26,
"learning_rate": 4.977563455402693e-05,
"loss": 1.4878,
"step": 190
},
{
"epoch": 0.27,
"learning_rate": 4.975143587010432e-05,
"loss": 1.5623,
"step": 200
},
{
"epoch": 0.28,
"learning_rate": 4.97260046830541e-05,
"loss": 1.6067,
"step": 210
},
{
"epoch": 0.3,
"learning_rate": 4.969934225922777e-05,
"loss": 1.4726,
"step": 220
},
{
"epoch": 0.31,
"learning_rate": 4.9671449926286494e-05,
"loss": 1.4386,
"step": 230
},
{
"epoch": 0.32,
"learning_rate": 4.964232907313507e-05,
"loss": 1.5058,
"step": 240
},
{
"epoch": 0.34,
"learning_rate": 4.9611981149852705e-05,
"loss": 1.4842,
"step": 250
},
{
"epoch": 0.35,
"learning_rate": 4.9580407667620835e-05,
"loss": 1.4038,
"step": 260
},
{
"epoch": 0.36,
"learning_rate": 4.9547610198647855e-05,
"loss": 1.4696,
"step": 270
},
{
"epoch": 0.38,
"learning_rate": 4.951359037609088e-05,
"loss": 1.5231,
"step": 280
},
{
"epoch": 0.39,
"learning_rate": 4.9478349893974345e-05,
"loss": 1.341,
"step": 290
},
{
"epoch": 0.4,
"learning_rate": 4.9441890507105735e-05,
"loss": 1.4268,
"step": 300
},
{
"epoch": 0.42,
"learning_rate": 4.9404214030988126e-05,
"loss": 1.4481,
"step": 310
},
{
"epoch": 0.43,
"learning_rate": 4.9365322341729845e-05,
"loss": 1.4532,
"step": 320
},
{
"epoch": 0.44,
"learning_rate": 4.9325217375951006e-05,
"loss": 1.2414,
"step": 330
},
{
"epoch": 0.46,
"learning_rate": 4.9283901130687094e-05,
"loss": 1.3546,
"step": 340
},
{
"epoch": 0.47,
"learning_rate": 4.9241375663289516e-05,
"loss": 1.3501,
"step": 350
},
{
"epoch": 0.49,
"learning_rate": 4.9197643091323165e-05,
"loss": 1.4564,
"step": 360
},
{
"epoch": 0.5,
"learning_rate": 4.9152705592460964e-05,
"loss": 1.4721,
"step": 370
},
{
"epoch": 0.51,
"learning_rate": 4.910656540437544e-05,
"loss": 1.5005,
"step": 380
},
{
"epoch": 0.53,
"learning_rate": 4.905922482462727e-05,
"loss": 1.3807,
"step": 390
},
{
"epoch": 0.54,
"learning_rate": 4.901068621055093e-05,
"loss": 1.3058,
"step": 400
},
{
"epoch": 0.55,
"learning_rate": 4.8960951979137236e-05,
"loss": 1.321,
"step": 410
},
{
"epoch": 0.57,
"learning_rate": 4.891002460691306e-05,
"loss": 1.4186,
"step": 420
},
{
"epoch": 0.58,
"learning_rate": 4.885790662981797e-05,
"loss": 1.5187,
"step": 430
},
{
"epoch": 0.59,
"learning_rate": 4.8804600643077954e-05,
"loss": 1.505,
"step": 440
},
{
"epoch": 0.61,
"learning_rate": 4.8750109301076205e-05,
"loss": 1.2945,
"step": 450
},
{
"epoch": 0.62,
"learning_rate": 4.869443531722094e-05,
"loss": 1.4216,
"step": 460
},
{
"epoch": 0.63,
"learning_rate": 4.863758146381029e-05,
"loss": 1.3838,
"step": 470
},
{
"epoch": 0.65,
"learning_rate": 4.8579550571894225e-05,
"loss": 1.2366,
"step": 480
},
{
"epoch": 0.66,
"learning_rate": 4.8520345531133636e-05,
"loss": 1.3815,
"step": 490
},
{
"epoch": 0.67,
"learning_rate": 4.845996928965638e-05,
"loss": 1.2919,
"step": 500
},
{
"epoch": 0.69,
"learning_rate": 4.839842485391053e-05,
"loss": 1.3449,
"step": 510
},
{
"epoch": 0.7,
"learning_rate": 4.833571528851463e-05,
"loss": 1.4683,
"step": 520
},
{
"epoch": 0.71,
"learning_rate": 4.827184371610511e-05,
"loss": 1.3436,
"step": 530
},
{
"epoch": 0.73,
"learning_rate": 4.820681331718081e-05,
"loss": 1.2934,
"step": 540
},
{
"epoch": 0.74,
"learning_rate": 4.8140627329944566e-05,
"loss": 1.2362,
"step": 550
},
{
"epoch": 0.75,
"learning_rate": 4.8073289050142015e-05,
"loss": 1.3885,
"step": 560
},
{
"epoch": 0.77,
"learning_rate": 4.80048018308974e-05,
"loss": 1.3071,
"step": 570
},
{
"epoch": 0.78,
"learning_rate": 4.7935169082546705e-05,
"loss": 1.3223,
"step": 580
},
{
"epoch": 0.8,
"learning_rate": 4.7864394272467764e-05,
"loss": 1.2082,
"step": 590
},
{
"epoch": 0.81,
"learning_rate": 4.7792480924907624e-05,
"loss": 1.202,
"step": 600
},
{
"epoch": 0.82,
"learning_rate": 4.7719432620807046e-05,
"loss": 1.2511,
"step": 610
},
{
"epoch": 0.84,
"learning_rate": 4.764525299762222e-05,
"loss": 1.1798,
"step": 620
},
{
"epoch": 0.85,
"learning_rate": 4.756994574914359e-05,
"loss": 1.3269,
"step": 630
},
{
"epoch": 0.86,
"learning_rate": 4.749351462531196e-05,
"loss": 1.4226,
"step": 640
},
{
"epoch": 0.88,
"learning_rate": 4.741596343203175e-05,
"loss": 1.3431,
"step": 650
},
{
"epoch": 0.89,
"learning_rate": 4.733729603098149e-05,
"loss": 1.2419,
"step": 660
},
{
"epoch": 0.9,
"learning_rate": 4.7257516339421494e-05,
"loss": 1.1961,
"step": 670
},
{
"epoch": 0.92,
"learning_rate": 4.7176628329998854e-05,
"loss": 1.4433,
"step": 680
},
{
"epoch": 0.93,
"learning_rate": 4.7094636030549564e-05,
"loss": 1.1532,
"step": 690
},
{
"epoch": 0.94,
"learning_rate": 4.7011543523897996e-05,
"loss": 1.2486,
"step": 700
},
{
"epoch": 0.96,
"learning_rate": 4.692735494765358e-05,
"loss": 1.2433,
"step": 710
},
{
"epoch": 0.97,
"learning_rate": 4.684207449400476e-05,
"loss": 1.3521,
"step": 720
},
{
"epoch": 0.98,
"learning_rate": 4.675570640951025e-05,
"loss": 1.264,
"step": 730
},
{
"epoch": 1.0,
"learning_rate": 4.666825499488759e-05,
"loss": 1.2021,
"step": 740
},
{
"epoch": 1.01,
"learning_rate": 4.657972460479899e-05,
"loss": 1.1499,
"step": 750
},
{
"epoch": 1.02,
"learning_rate": 4.649011964763444e-05,
"loss": 1.2565,
"step": 760
},
{
"epoch": 1.04,
"learning_rate": 4.639944458529226e-05,
"loss": 1.2917,
"step": 770
},
{
"epoch": 1.05,
"learning_rate": 4.630770393295688e-05,
"loss": 1.3944,
"step": 780
},
{
"epoch": 1.06,
"learning_rate": 4.6214902258874034e-05,
"loss": 1.3155,
"step": 790
},
{
"epoch": 1.08,
"learning_rate": 4.612104418412323e-05,
"loss": 1.1417,
"step": 800
},
{
"epoch": 1.09,
"learning_rate": 4.60261343823877e-05,
"loss": 1.3151,
"step": 810
},
{
"epoch": 1.11,
"learning_rate": 4.593017757972165e-05,
"loss": 1.3504,
"step": 820
},
{
"epoch": 1.12,
"learning_rate": 4.5833178554314915e-05,
"loss": 1.1928,
"step": 830
},
{
"epoch": 1.13,
"learning_rate": 4.573514213625505e-05,
"loss": 1.1215,
"step": 840
},
{
"epoch": 1.15,
"learning_rate": 4.563607320728679e-05,
"loss": 1.3223,
"step": 850
},
{
"epoch": 1.16,
"learning_rate": 4.553597670056897e-05,
"loss": 1.1495,
"step": 860
},
{
"epoch": 1.17,
"learning_rate": 4.543485760042892e-05,
"loss": 1.2296,
"step": 870
},
{
"epoch": 1.19,
"learning_rate": 4.533272094211418e-05,
"loss": 1.1743,
"step": 880
},
{
"epoch": 1.2,
"learning_rate": 4.522957181154185e-05,
"loss": 1.2596,
"step": 890
},
{
"epoch": 1.21,
"learning_rate": 4.5125415345045316e-05,
"loss": 1.3216,
"step": 900
},
{
"epoch": 1.23,
"learning_rate": 4.5020256729118446e-05,
"loss": 1.2431,
"step": 910
},
{
"epoch": 1.24,
"learning_rate": 4.491410120015737e-05,
"loss": 1.2134,
"step": 920
},
{
"epoch": 1.25,
"learning_rate": 4.4806954044199725e-05,
"loss": 1.3169,
"step": 930
},
{
"epoch": 1.27,
"learning_rate": 4.469882059666142e-05,
"loss": 1.1074,
"step": 940
},
{
"epoch": 1.28,
"learning_rate": 4.458970624207098e-05,
"loss": 1.2713,
"step": 950
},
{
"epoch": 1.29,
"learning_rate": 4.4479616413801396e-05,
"loss": 1.2656,
"step": 960
},
{
"epoch": 1.31,
"learning_rate": 4.4368556593799593e-05,
"loss": 1.2994,
"step": 970
},
{
"epoch": 1.32,
"learning_rate": 4.425653231231344e-05,
"loss": 1.3854,
"step": 980
},
{
"epoch": 1.33,
"learning_rate": 4.414354914761638e-05,
"loss": 1.2299,
"step": 990
},
{
"epoch": 1.35,
"learning_rate": 4.402961272572964e-05,
"loss": 1.3386,
"step": 1000
},
{
"epoch": 1.36,
"learning_rate": 4.391472872014212e-05,
"loss": 1.2417,
"step": 1010
},
{
"epoch": 1.37,
"learning_rate": 4.379890285152782e-05,
"loss": 1.3562,
"step": 1020
},
{
"epoch": 1.39,
"learning_rate": 4.368214088746107e-05,
"loss": 1.0887,
"step": 1030
},
{
"epoch": 1.4,
"learning_rate": 4.356444864212921e-05,
"loss": 1.2619,
"step": 1040
},
{
"epoch": 1.42,
"learning_rate": 4.344583197604318e-05,
"loss": 1.1281,
"step": 1050
},
{
"epoch": 1.43,
"learning_rate": 4.332629679574566e-05,
"loss": 1.123,
"step": 1060
},
{
"epoch": 1.44,
"learning_rate": 4.320584905351691e-05,
"loss": 1.2429,
"step": 1070
},
{
"epoch": 1.46,
"learning_rate": 4.308449474707843e-05,
"loss": 1.1914,
"step": 1080
},
{
"epoch": 1.47,
"learning_rate": 4.296223991929429e-05,
"loss": 1.1858,
"step": 1090
},
{
"epoch": 1.48,
"learning_rate": 4.2839090657870196e-05,
"loss": 1.1781,
"step": 1100
},
{
"epoch": 1.5,
"learning_rate": 4.271505309505038e-05,
"loss": 1.2164,
"step": 1110
},
{
"epoch": 1.51,
"learning_rate": 4.259013340731224e-05,
"loss": 1.0746,
"step": 1120
},
{
"epoch": 1.52,
"learning_rate": 4.246433781505876e-05,
"loss": 1.0874,
"step": 1130
},
{
"epoch": 1.54,
"learning_rate": 4.2337672582308815e-05,
"loss": 1.1537,
"step": 1140
},
{
"epoch": 1.55,
"learning_rate": 4.221014401638517e-05,
"loss": 1.3942,
"step": 1150
},
{
"epoch": 1.56,
"learning_rate": 4.20817584676005e-05,
"loss": 1.0343,
"step": 1160
},
{
"epoch": 1.58,
"learning_rate": 4.195252232894108e-05,
"loss": 1.2696,
"step": 1170
},
{
"epoch": 1.59,
"learning_rate": 4.1822442035748534e-05,
"loss": 1.2511,
"step": 1180
},
{
"epoch": 1.6,
"learning_rate": 4.1691524065399324e-05,
"loss": 1.0978,
"step": 1190
},
{
"epoch": 1.62,
"learning_rate": 4.155977493698222e-05,
"loss": 1.1222,
"step": 1200
},
{
"epoch": 1.63,
"learning_rate": 4.142720121097372e-05,
"loss": 1.1663,
"step": 1210
},
{
"epoch": 1.64,
"learning_rate": 4.129380948891132e-05,
"loss": 1.209,
"step": 1220
},
{
"epoch": 1.66,
"learning_rate": 4.1159606413064786e-05,
"loss": 1.2177,
"step": 1230
},
{
"epoch": 1.67,
"learning_rate": 4.1024598666105444e-05,
"loss": 1.1437,
"step": 1240
},
{
"epoch": 1.68,
"learning_rate": 4.088879297077341e-05,
"loss": 1.0417,
"step": 1250
},
{
"epoch": 1.7,
"learning_rate": 4.075219608954278e-05,
"loss": 1.1635,
"step": 1260
},
{
"epoch": 1.71,
"learning_rate": 4.0614814824284943e-05,
"loss": 1.0666,
"step": 1270
},
{
"epoch": 1.73,
"learning_rate": 4.047665601592984e-05,
"loss": 1.3199,
"step": 1280
},
{
"epoch": 1.74,
"learning_rate": 4.0337726544125376e-05,
"loss": 1.1025,
"step": 1290
},
{
"epoch": 1.75,
"learning_rate": 4.019803332689478e-05,
"loss": 0.9055,
"step": 1300
},
{
"epoch": 1.77,
"learning_rate": 4.0057583320292154e-05,
"loss": 1.0077,
"step": 1310
},
{
"epoch": 1.78,
"learning_rate": 3.991638351805611e-05,
"loss": 1.1106,
"step": 1320
},
{
"epoch": 1.79,
"learning_rate": 3.977444095126146e-05,
"loss": 1.0175,
"step": 1330
},
{
"epoch": 1.81,
"learning_rate": 3.9631762687969184e-05,
"loss": 1.1105,
"step": 1340
},
{
"epoch": 1.82,
"learning_rate": 3.948835583287439e-05,
"loss": 0.96,
"step": 1350
},
{
"epoch": 1.83,
"learning_rate": 3.93442275269526e-05,
"loss": 1.2329,
"step": 1360
},
{
"epoch": 1.85,
"learning_rate": 3.9199384947104136e-05,
"loss": 1.0752,
"step": 1370
},
{
"epoch": 1.86,
"learning_rate": 3.9053835305796706e-05,
"loss": 0.9997,
"step": 1380
},
{
"epoch": 1.87,
"learning_rate": 3.890758585070635e-05,
"loss": 1.1088,
"step": 1390
},
{
"epoch": 1.89,
"learning_rate": 3.876064386435646e-05,
"loss": 0.94,
"step": 1400
},
{
"epoch": 1.9,
"learning_rate": 3.861301666375517e-05,
"loss": 1.1949,
"step": 1410
},
{
"epoch": 1.91,
"learning_rate": 3.846471160003102e-05,
"loss": 1.2521,
"step": 1420
},
{
"epoch": 1.93,
"learning_rate": 3.831573605806689e-05,
"loss": 1.1376,
"step": 1430
},
{
"epoch": 1.94,
"learning_rate": 3.816609745613224e-05,
"loss": 1.1883,
"step": 1440
},
{
"epoch": 1.95,
"learning_rate": 3.80158032455138e-05,
"loss": 1.3168,
"step": 1450
},
{
"epoch": 1.97,
"learning_rate": 3.7864860910144425e-05,
"loss": 1.0501,
"step": 1460
},
{
"epoch": 1.98,
"learning_rate": 3.7713277966230514e-05,
"loss": 1.1886,
"step": 1470
},
{
"epoch": 1.99,
"learning_rate": 3.7561061961877694e-05,
"loss": 1.1367,
"step": 1480
},
{
"epoch": 2.01,
"learning_rate": 3.740822047671498e-05,
"loss": 1.1561,
"step": 1490
},
{
"epoch": 2.02,
"learning_rate": 3.725476112151735e-05,
"loss": 1.0747,
"step": 1500
},
{
"epoch": 2.04,
"learning_rate": 3.710069153782672e-05,
"loss": 1.0443,
"step": 1510
},
{
"epoch": 2.05,
"learning_rate": 3.6946019397571496e-05,
"loss": 1.3622,
"step": 1520
},
{
"epoch": 2.06,
"learning_rate": 3.679075240268452e-05,
"loss": 1.1274,
"step": 1530
},
{
"epoch": 2.08,
"learning_rate": 3.663489828471953e-05,
"loss": 0.9682,
"step": 1540
},
{
"epoch": 2.09,
"learning_rate": 3.647846480446621e-05,
"loss": 0.9268,
"step": 1550
},
{
"epoch": 2.1,
"learning_rate": 3.6321459751563703e-05,
"loss": 1.1239,
"step": 1560
},
{
"epoch": 2.12,
"learning_rate": 3.6163890944112715e-05,
"loss": 1.3213,
"step": 1570
},
{
"epoch": 2.13,
"learning_rate": 3.600576622828627e-05,
"loss": 1.1064,
"step": 1580
},
{
"epoch": 2.14,
"learning_rate": 3.5847093477938956e-05,
"loss": 1.2662,
"step": 1590
},
{
"epoch": 2.16,
"learning_rate": 3.568788059421484e-05,
"loss": 1.0053,
"step": 1600
},
{
"epoch": 2.17,
"learning_rate": 3.5528135505154084e-05,
"loss": 1.0666,
"step": 1610
},
{
"epoch": 2.18,
"learning_rate": 3.5367866165298084e-05,
"loss": 1.1048,
"step": 1620
},
{
"epoch": 2.2,
"learning_rate": 3.520708055529347e-05,
"loss": 1.0777,
"step": 1630
},
{
"epoch": 2.21,
"learning_rate": 3.504578668149462e-05,
"loss": 1.0932,
"step": 1640
},
{
"epoch": 2.22,
"learning_rate": 3.488399257556502e-05,
"loss": 1.0945,
"step": 1650
},
{
"epoch": 2.24,
"learning_rate": 3.472170629407735e-05,
"loss": 1.1311,
"step": 1660
},
{
"epoch": 2.25,
"learning_rate": 3.455893591811225e-05,
"loss": 1.2245,
"step": 1670
},
{
"epoch": 2.26,
"learning_rate": 3.4395689552855955e-05,
"loss": 1.088,
"step": 1680
},
{
"epoch": 2.28,
"learning_rate": 3.423197532719671e-05,
"loss": 0.9423,
"step": 1690
},
{
"epoch": 2.29,
"learning_rate": 3.4067801393319954e-05,
"loss": 0.9905,
"step": 1700
},
{
"epoch": 2.3,
"learning_rate": 3.390317592630239e-05,
"loss": 0.9556,
"step": 1710
},
{
"epoch": 2.32,
"learning_rate": 3.3738107123704934e-05,
"loss": 1.1236,
"step": 1720
},
{
"epoch": 2.33,
"learning_rate": 3.357260320516447e-05,
"loss": 0.8063,
"step": 1730
},
{
"epoch": 2.35,
"learning_rate": 3.34066724119846e-05,
"loss": 1.0238,
"step": 1740
},
{
"epoch": 2.36,
"learning_rate": 3.32403230067252e-05,
"loss": 0.8993,
"step": 1750
},
{
"epoch": 2.37,
"learning_rate": 3.3073563272791074e-05,
"loss": 1.0067,
"step": 1760
},
{
"epoch": 2.39,
"learning_rate": 3.2906401514019406e-05,
"loss": 1.094,
"step": 1770
},
{
"epoch": 2.4,
"learning_rate": 3.2738846054266296e-05,
"loss": 1.1968,
"step": 1780
},
{
"epoch": 2.41,
"learning_rate": 3.2570905236992274e-05,
"loss": 1.0276,
"step": 1790
},
{
"epoch": 2.43,
"learning_rate": 3.240258742484683e-05,
"loss": 1.0161,
"step": 1800
},
{
"epoch": 2.44,
"learning_rate": 3.2233900999251995e-05,
"loss": 1.1456,
"step": 1810
},
{
"epoch": 2.45,
"learning_rate": 3.206485435998498e-05,
"loss": 0.9577,
"step": 1820
},
{
"epoch": 2.47,
"learning_rate": 3.189545592475992e-05,
"loss": 1.0474,
"step": 1830
},
{
"epoch": 2.48,
"learning_rate": 3.172571412880872e-05,
"loss": 1.1945,
"step": 1840
},
{
"epoch": 2.49,
"learning_rate": 3.1555637424460984e-05,
"loss": 0.994,
"step": 1850
},
{
"epoch": 2.51,
"learning_rate": 3.138523428072316e-05,
"loss": 0.9061,
"step": 1860
},
{
"epoch": 2.52,
"learning_rate": 3.121451318285684e-05,
"loss": 0.9634,
"step": 1870
},
{
"epoch": 2.53,
"learning_rate": 3.104348263195617e-05,
"loss": 1.2188,
"step": 1880
},
{
"epoch": 2.55,
"learning_rate": 3.0872151144524595e-05,
"loss": 0.9831,
"step": 1890
},
{
"epoch": 2.56,
"learning_rate": 3.070052725205077e-05,
"loss": 1.0728,
"step": 1900
},
{
"epoch": 2.57,
"learning_rate": 3.052861950058371e-05,
"loss": 1.0418,
"step": 1910
},
{
"epoch": 2.59,
"learning_rate": 3.0356436450307256e-05,
"loss": 1.2411,
"step": 1920
},
{
"epoch": 2.6,
"learning_rate": 3.0183986675113814e-05,
"loss": 0.8613,
"step": 1930
},
{
"epoch": 2.61,
"learning_rate": 3.0011278762177415e-05,
"loss": 1.0478,
"step": 1940
},
{
"epoch": 2.63,
"learning_rate": 2.9838321311526107e-05,
"loss": 0.82,
"step": 1950
},
{
"epoch": 2.64,
"learning_rate": 2.9665122935613727e-05,
"loss": 0.9665,
"step": 1960
},
{
"epoch": 2.65,
"learning_rate": 2.9491692258891045e-05,
"loss": 0.9478,
"step": 1970
},
{
"epoch": 2.67,
"learning_rate": 2.9318037917376297e-05,
"loss": 0.8923,
"step": 1980
},
{
"epoch": 2.68,
"learning_rate": 2.9144168558225138e-05,
"loss": 0.9888,
"step": 1990
},
{
"epoch": 2.7,
"learning_rate": 2.8970092839300106e-05,
"loss": 0.7904,
"step": 2000
},
{
"epoch": 2.71,
"learning_rate": 2.879581942873945e-05,
"loss": 0.8784,
"step": 2010
},
{
"epoch": 2.72,
"learning_rate": 2.8621357004525506e-05,
"loss": 1.0034,
"step": 2020
},
{
"epoch": 2.74,
"learning_rate": 2.8446714254052613e-05,
"loss": 1.0585,
"step": 2030
},
{
"epoch": 2.75,
"learning_rate": 2.827189987369446e-05,
"loss": 1.0363,
"step": 2040
},
{
"epoch": 2.76,
"learning_rate": 2.809692256837113e-05,
"loss": 0.8699,
"step": 2050
},
{
"epoch": 2.78,
"learning_rate": 2.7921791051115548e-05,
"loss": 0.9598,
"step": 2060
},
{
"epoch": 2.79,
"learning_rate": 2.7746514042639677e-05,
"loss": 0.8898,
"step": 2070
},
{
"epoch": 2.8,
"learning_rate": 2.757110027090023e-05,
"loss": 1.0903,
"step": 2080
},
{
"epoch": 2.82,
"learning_rate": 2.7395558470664097e-05,
"loss": 0.9971,
"step": 2090
},
{
"epoch": 2.83,
"learning_rate": 2.7219897383073373e-05,
"loss": 1.103,
"step": 2100
},
{
"epoch": 2.84,
"learning_rate": 2.704412575521007e-05,
"loss": 1.0361,
"step": 2110
},
{
"epoch": 2.86,
"learning_rate": 2.686825233966061e-05,
"loss": 1.1389,
"step": 2120
},
{
"epoch": 2.87,
"learning_rate": 2.6692285894079923e-05,
"loss": 1.0003,
"step": 2130
},
{
"epoch": 2.88,
"learning_rate": 2.6516235180755412e-05,
"loss": 0.9947,
"step": 2140
},
{
"epoch": 2.9,
"learning_rate": 2.6340108966170592e-05,
"loss": 0.8088,
"step": 2150
},
{
"epoch": 2.91,
"learning_rate": 2.6163916020568595e-05,
"loss": 0.7742,
"step": 2160
},
{
"epoch": 2.92,
"learning_rate": 2.5987665117515446e-05,
"loss": 1.0853,
"step": 2170
},
{
"epoch": 2.94,
"learning_rate": 2.5811365033463147e-05,
"loss": 1.0034,
"step": 2180
},
{
"epoch": 2.95,
"learning_rate": 2.5635024547312704e-05,
"loss": 1.0611,
"step": 2190
},
{
"epoch": 2.96,
"learning_rate": 2.5458652439976932e-05,
"loss": 1.0789,
"step": 2200
},
{
"epoch": 2.98,
"learning_rate": 2.528225749394325e-05,
"loss": 1.1031,
"step": 2210
},
{
"epoch": 2.99,
"learning_rate": 2.510584849283632e-05,
"loss": 0.9604,
"step": 2220
},
{
"epoch": 3.01,
"learning_rate": 2.492943422098069e-05,
"loss": 0.992,
"step": 2230
},
{
"epoch": 3.02,
"learning_rate": 2.475302346296336e-05,
"loss": 0.9957,
"step": 2240
},
{
"epoch": 3.03,
"learning_rate": 2.4576625003196356e-05,
"loss": 0.9671,
"step": 2250
},
{
"epoch": 3.05,
"learning_rate": 2.4400247625479326e-05,
"loss": 0.9436,
"step": 2260
},
{
"epoch": 3.06,
"learning_rate": 2.4223900112562107e-05,
"loss": 1.0099,
"step": 2270
},
{
"epoch": 3.07,
"learning_rate": 2.4047591245707418e-05,
"loss": 0.8495,
"step": 2280
},
{
"epoch": 3.09,
"learning_rate": 2.387132980425362e-05,
"loss": 0.9465,
"step": 2290
},
{
"epoch": 3.1,
"learning_rate": 2.3695124565177467e-05,
"loss": 1.1073,
"step": 2300
},
{
"epoch": 3.11,
"learning_rate": 2.3518984302657146e-05,
"loss": 1.0246,
"step": 2310
},
{
"epoch": 3.13,
"learning_rate": 2.33429177876353e-05,
"loss": 0.913,
"step": 2320
},
{
"epoch": 3.14,
"learning_rate": 2.3166933787382292e-05,
"loss": 1.0195,
"step": 2330
},
{
"epoch": 3.15,
"learning_rate": 2.2991041065059672e-05,
"loss": 1.009,
"step": 2340
},
{
"epoch": 3.17,
"learning_rate": 2.2815248379283745e-05,
"loss": 0.9388,
"step": 2350
},
{
"epoch": 3.18,
"learning_rate": 2.2639564483689532e-05,
"loss": 0.7509,
"step": 2360
},
{
"epoch": 3.19,
"learning_rate": 2.2463998126494764e-05,
"loss": 0.9865,
"step": 2370
},
{
"epoch": 3.21,
"learning_rate": 2.2288558050064367e-05,
"loss": 1.0535,
"step": 2380
},
{
"epoch": 3.22,
"learning_rate": 2.2113252990475056e-05,
"loss": 0.8899,
"step": 2390
},
{
"epoch": 3.23,
"learning_rate": 2.1938091677080365e-05,
"loss": 1.0849,
"step": 2400
},
{
"epoch": 3.25,
"learning_rate": 2.176308283207596e-05,
"loss": 1.0134,
"step": 2410
},
{
"epoch": 3.26,
"learning_rate": 2.158823517006532e-05,
"loss": 0.9374,
"step": 2420
},
{
"epoch": 3.27,
"learning_rate": 2.141355739762576e-05,
"loss": 0.769,
"step": 2430
},
{
"epoch": 3.29,
"learning_rate": 2.1239058212874904e-05,
"loss": 0.8231,
"step": 2440
},
{
"epoch": 3.3,
"learning_rate": 2.1064746305037595e-05,
"loss": 0.9953,
"step": 2450
},
{
"epoch": 3.32,
"learning_rate": 2.0890630354013153e-05,
"loss": 1.161,
"step": 2460
},
{
"epoch": 3.33,
"learning_rate": 2.0716719029943184e-05,
"loss": 0.8749,
"step": 2470
},
{
"epoch": 3.34,
"learning_rate": 2.0543020992779868e-05,
"loss": 0.8434,
"step": 2480
},
{
"epoch": 3.36,
"learning_rate": 2.036954489185472e-05,
"loss": 0.9605,
"step": 2490
},
{
"epoch": 3.37,
"learning_rate": 2.0196299365447878e-05,
"loss": 0.9528,
"step": 2500
},
{
"epoch": 3.38,
"learning_rate": 2.0023293040357965e-05,
"loss": 0.779,
"step": 2510
},
{
"epoch": 3.4,
"learning_rate": 1.9850534531472546e-05,
"loss": 0.9913,
"step": 2520
},
{
"epoch": 3.41,
"learning_rate": 1.9678032441339102e-05,
"loss": 1.0698,
"step": 2530
},
{
"epoch": 3.42,
"learning_rate": 1.9505795359736694e-05,
"loss": 0.9707,
"step": 2540
},
{
"epoch": 3.44,
"learning_rate": 1.9333831863248227e-05,
"loss": 0.9474,
"step": 2550
},
{
"epoch": 3.45,
"learning_rate": 1.916215051483339e-05,
"loss": 0.8685,
"step": 2560
},
{
"epoch": 3.46,
"learning_rate": 1.8990759863402217e-05,
"loss": 0.9708,
"step": 2570
},
{
"epoch": 3.48,
"learning_rate": 1.881966844338945e-05,
"loss": 0.9904,
"step": 2580
},
{
"epoch": 3.49,
"learning_rate": 1.8648884774329524e-05,
"loss": 0.9169,
"step": 2590
},
{
"epoch": 3.5,
"learning_rate": 1.8478417360432352e-05,
"loss": 0.9897,
"step": 2600
},
{
"epoch": 3.52,
"learning_rate": 1.830827469015984e-05,
"loss": 0.9619,
"step": 2610
},
{
"epoch": 3.53,
"learning_rate": 1.8138465235803232e-05,
"loss": 0.8307,
"step": 2620
},
{
"epoch": 3.54,
"learning_rate": 1.7968997453061222e-05,
"loss": 0.8884,
"step": 2630
},
{
"epoch": 3.56,
"learning_rate": 1.779987978061886e-05,
"loss": 0.9401,
"step": 2640
},
{
"epoch": 3.57,
"learning_rate": 1.7631120639727393e-05,
"loss": 0.7523,
"step": 2650
},
{
"epoch": 3.58,
"learning_rate": 1.746272843378493e-05,
"loss": 0.896,
"step": 2660
},
{
"epoch": 3.6,
"learning_rate": 1.7294711547917942e-05,
"loss": 0.8265,
"step": 2670
},
{
"epoch": 3.61,
"learning_rate": 1.712707834856377e-05,
"loss": 0.9221,
"step": 2680
},
{
"epoch": 3.63,
"learning_rate": 1.6959837183054e-05,
"loss": 0.7205,
"step": 2690
},
{
"epoch": 3.64,
"learning_rate": 1.6792996379198828e-05,
"loss": 0.9376,
"step": 2700
},
{
"epoch": 3.65,
"learning_rate": 1.662656424487233e-05,
"loss": 0.988,
"step": 2710
},
{
"epoch": 3.67,
"learning_rate": 1.6460549067598782e-05,
"loss": 0.8103,
"step": 2720
},
{
"epoch": 3.68,
"learning_rate": 1.6294959114140034e-05,
"loss": 0.9497,
"step": 2730
},
{
"epoch": 3.69,
"learning_rate": 1.612980263008379e-05,
"loss": 0.8883,
"step": 2740
},
{
"epoch": 3.71,
"learning_rate": 1.5965087839433054e-05,
"loss": 1.053,
"step": 2750
},
{
"epoch": 3.72,
"learning_rate": 1.580082294419664e-05,
"loss": 1.0284,
"step": 2760
},
{
"epoch": 3.73,
"learning_rate": 1.563701612398068e-05,
"loss": 1.0539,
"step": 2770
},
{
"epoch": 3.75,
"learning_rate": 1.54736755355814e-05,
"loss": 0.8497,
"step": 2780
},
{
"epoch": 3.76,
"learning_rate": 1.5310809312578875e-05,
"loss": 0.9941,
"step": 2790
},
{
"epoch": 3.77,
"learning_rate": 1.5148425564932084e-05,
"loss": 1.0088,
"step": 2800
},
{
"epoch": 3.79,
"learning_rate": 1.4986532378575025e-05,
"loss": 1.0044,
"step": 2810
},
{
"epoch": 3.8,
"learning_rate": 1.4825137815014073e-05,
"loss": 0.9474,
"step": 2820
},
{
"epoch": 3.81,
"learning_rate": 1.466424991092662e-05,
"loss": 1.0328,
"step": 2830
},
{
"epoch": 3.83,
"learning_rate": 1.4503876677760797e-05,
"loss": 1.0608,
"step": 2840
},
{
"epoch": 3.84,
"learning_rate": 1.4344026101336594e-05,
"loss": 0.8027,
"step": 2850
},
{
"epoch": 3.85,
"learning_rate": 1.4184706141448207e-05,
"loss": 0.7189,
"step": 2860
},
{
"epoch": 3.87,
"learning_rate": 1.402592473146766e-05,
"loss": 0.8236,
"step": 2870
},
{
"epoch": 3.88,
"learning_rate": 1.3867689777949754e-05,
"loss": 0.9332,
"step": 2880
},
{
"epoch": 3.89,
"learning_rate": 1.3710009160238385e-05,
"loss": 0.8502,
"step": 2890
},
{
"epoch": 3.91,
"learning_rate": 1.35528907300742e-05,
"loss": 1.0084,
"step": 2900
},
{
"epoch": 3.92,
"learning_rate": 1.3396342311203541e-05,
"loss": 1.1641,
"step": 2910
},
{
"epoch": 3.94,
"learning_rate": 1.3240371698988957e-05,
"loss": 0.8071,
"step": 2920
},
{
"epoch": 3.95,
"learning_rate": 1.3084986660020984e-05,
"loss": 0.6047,
"step": 2930
},
{
"epoch": 3.96,
"learning_rate": 1.2930194931731382e-05,
"loss": 0.8257,
"step": 2940
},
{
"epoch": 3.98,
"learning_rate": 1.2776004222007908e-05,
"loss": 1.0769,
"step": 2950
},
{
"epoch": 3.99,
"learning_rate": 1.2622422208810455e-05,
"loss": 1.0479,
"step": 2960
},
{
"epoch": 4.0,
"learning_rate": 1.2469456539788748e-05,
"loss": 0.9407,
"step": 2970
},
{
"epoch": 4.02,
"learning_rate": 1.2317114831901521e-05,
"loss": 0.9241,
"step": 2980
},
{
"epoch": 4.03,
"learning_rate": 1.2165404671037205e-05,
"loss": 0.8023,
"step": 2990
},
{
"epoch": 4.04,
"learning_rate": 1.2014333611636238e-05,
"loss": 1.0193,
"step": 3000
},
{
"epoch": 4.06,
"learning_rate": 1.1863909176314856e-05,
"loss": 0.9487,
"step": 3010
},
{
"epoch": 4.07,
"learning_rate": 1.1714138855490484e-05,
"loss": 0.7876,
"step": 3020
},
{
"epoch": 4.08,
"learning_rate": 1.1565030107008803e-05,
"loss": 1.0967,
"step": 3030
},
{
"epoch": 4.1,
"learning_rate": 1.1416590355772347e-05,
"loss": 1.0072,
"step": 3040
},
{
"epoch": 4.11,
"learning_rate": 1.1268826993370781e-05,
"loss": 1.0347,
"step": 3050
},
{
"epoch": 4.12,
"learning_rate": 1.1121747377712858e-05,
"loss": 0.9914,
"step": 3060
},
{
"epoch": 4.14,
"learning_rate": 1.0975358832659977e-05,
"loss": 1.0746,
"step": 3070
},
{
"epoch": 4.15,
"learning_rate": 1.0829668647661559e-05,
"loss": 0.9422,
"step": 3080
},
{
"epoch": 4.16,
"learning_rate": 1.0684684077392027e-05,
"loss": 1.0835,
"step": 3090
},
{
"epoch": 4.18,
"learning_rate": 1.0540412341389561e-05,
"loss": 0.717,
"step": 3100
},
{
"epoch": 4.19,
"learning_rate": 1.0396860623696603e-05,
"loss": 0.8606,
"step": 3110
},
{
"epoch": 4.2,
"learning_rate": 1.0254036072502141e-05,
"loss": 0.8979,
"step": 3120
},
{
"epoch": 4.22,
"learning_rate": 1.0111945799785719e-05,
"loss": 1.0647,
"step": 3130
},
{
"epoch": 4.23,
"learning_rate": 9.97059688096335e-06,
"loss": 0.8494,
"step": 3140
},
{
"epoch": 4.25,
"learning_rate": 9.829996354535172e-06,
"loss": 0.7145,
"step": 3150
},
{
"epoch": 4.26,
"learning_rate": 9.690151221734928e-06,
"loss": 0.8389,
"step": 3160
},
{
"epoch": 4.27,
"learning_rate": 9.551068446181422e-06,
"loss": 0.8896,
"step": 3170
},
{
"epoch": 4.29,
"learning_rate": 9.412754953531663e-06,
"loss": 0.9955,
"step": 3180
},
{
"epoch": 4.3,
"learning_rate": 9.275217631136082e-06,
"loss": 0.9124,
"step": 3190
},
{
"epoch": 4.31,
"learning_rate": 9.138463327695542e-06,
"loss": 1.0565,
"step": 3200
},
{
"epoch": 4.33,
"learning_rate": 9.002498852920282e-06,
"loss": 0.8348,
"step": 3210
},
{
"epoch": 4.34,
"learning_rate": 8.867330977190877e-06,
"loss": 0.8811,
"step": 3220
},
{
"epoch": 4.35,
"learning_rate": 8.732966431221068e-06,
"loss": 0.8473,
"step": 3230
},
{
"epoch": 4.37,
"learning_rate": 8.599411905722626e-06,
"loss": 0.7923,
"step": 3240
},
{
"epoch": 4.38,
"learning_rate": 8.466674051072162e-06,
"loss": 0.8181,
"step": 3250
},
{
"epoch": 4.39,
"learning_rate": 8.334759476980004e-06,
"loss": 0.7708,
"step": 3260
},
{
"epoch": 4.41,
"learning_rate": 8.203674752161017e-06,
"loss": 0.8192,
"step": 3270
},
{
"epoch": 4.42,
"learning_rate": 8.073426404007568e-06,
"loss": 0.8737,
"step": 3280
},
{
"epoch": 4.43,
"learning_rate": 7.944020918264458e-06,
"loss": 0.8013,
"step": 3290
},
{
"epoch": 4.45,
"learning_rate": 7.815464738705953e-06,
"loss": 0.7705,
"step": 3300
},
{
"epoch": 4.46,
"learning_rate": 7.687764266814973e-06,
"loss": 0.7375,
"step": 3310
},
{
"epoch": 4.47,
"learning_rate": 7.5609258614642465e-06,
"loss": 0.7956,
"step": 3320
},
{
"epoch": 4.49,
"learning_rate": 7.434955838599738e-06,
"loss": 0.9865,
"step": 3330
},
{
"epoch": 4.5,
"learning_rate": 7.309860470926114e-06,
"loss": 0.8184,
"step": 3340
},
{
"epoch": 4.51,
"learning_rate": 7.185645987594378e-06,
"loss": 0.8872,
"step": 3350
},
{
"epoch": 4.53,
"learning_rate": 7.062318573891719e-06,
"loss": 1.108,
"step": 3360
},
{
"epoch": 4.54,
"learning_rate": 6.939884370933519e-06,
"loss": 0.9655,
"step": 3370
},
{
"epoch": 4.56,
"learning_rate": 6.818349475357505e-06,
"loss": 0.8781,
"step": 3380
},
{
"epoch": 4.57,
"learning_rate": 6.697719939020211e-06,
"loss": 0.7701,
"step": 3390
},
{
"epoch": 4.58,
"learning_rate": 6.578001768695624e-06,
"loss": 0.9079,
"step": 3400
},
{
"epoch": 4.6,
"learning_rate": 6.459200925776035e-06,
"loss": 0.8282,
"step": 3410
},
{
"epoch": 4.61,
"learning_rate": 6.341323325975238e-06,
"loss": 0.8578,
"step": 3420
},
{
"epoch": 4.62,
"learning_rate": 6.224374839033928e-06,
"loss": 1.0168,
"step": 3430
},
{
"epoch": 4.64,
"learning_rate": 6.108361288427431e-06,
"loss": 1.2925,
"step": 3440
},
{
"epoch": 4.65,
"learning_rate": 5.99328845107571e-06,
"loss": 0.928,
"step": 3450
},
{
"epoch": 4.66,
"learning_rate": 5.879162057055695e-06,
"loss": 0.9106,
"step": 3460
},
{
"epoch": 4.68,
"learning_rate": 5.76598778931598e-06,
"loss": 0.7693,
"step": 3470
},
{
"epoch": 4.69,
"learning_rate": 5.653771283393833e-06,
"loss": 0.8638,
"step": 3480
},
{
"epoch": 4.7,
"learning_rate": 5.54251812713453e-06,
"loss": 0.8289,
"step": 3490
},
{
"epoch": 4.72,
"learning_rate": 5.4322338604131715e-06,
"loss": 0.7716,
"step": 3500
},
{
"epoch": 4.73,
"learning_rate": 5.322923974858784e-06,
"loss": 0.7649,
"step": 3510
},
{
"epoch": 4.74,
"learning_rate": 5.21459391358087e-06,
"loss": 0.6982,
"step": 3520
},
{
"epoch": 4.76,
"learning_rate": 5.107249070898376e-06,
"loss": 0.8857,
"step": 3530
},
{
"epoch": 4.77,
"learning_rate": 5.00089479207107e-06,
"loss": 1.0044,
"step": 3540
},
{
"epoch": 4.78,
"learning_rate": 4.895536373033366e-06,
"loss": 1.0828,
"step": 3550
},
{
"epoch": 4.8,
"learning_rate": 4.79117906013064e-06,
"loss": 0.8075,
"step": 3560
},
{
"epoch": 4.81,
"learning_rate": 4.687828049857967e-06,
"loss": 0.9403,
"step": 3570
},
{
"epoch": 4.82,
"learning_rate": 4.585488488601358e-06,
"loss": 0.6863,
"step": 3580
},
{
"epoch": 4.84,
"learning_rate": 4.484165472381516e-06,
"loss": 0.8102,
"step": 3590
},
{
"epoch": 4.85,
"learning_rate": 4.38386404660003e-06,
"loss": 0.8119,
"step": 3600
},
{
"epoch": 4.87,
"learning_rate": 4.284589205788206e-06,
"loss": 0.8888,
"step": 3610
},
{
"epoch": 4.88,
"learning_rate": 4.186345893358315e-06,
"loss": 0.8364,
"step": 3620
},
{
"epoch": 4.89,
"learning_rate": 4.089139001357439e-06,
"loss": 0.8115,
"step": 3630
},
{
"epoch": 4.91,
"learning_rate": 3.992973370223896e-06,
"loss": 1.0722,
"step": 3640
},
{
"epoch": 4.92,
"learning_rate": 3.89785378854618e-06,
"loss": 1.0983,
"step": 3650
},
{
"epoch": 4.93,
"learning_rate": 3.803784992824533e-06,
"loss": 0.9268,
"step": 3660
},
{
"epoch": 4.95,
"learning_rate": 3.710771667235083e-06,
"loss": 0.8712,
"step": 3670
},
{
"epoch": 4.96,
"learning_rate": 3.618818443396579e-06,
"loss": 0.6253,
"step": 3680
},
{
"epoch": 4.97,
"learning_rate": 3.52792990013979e-06,
"loss": 0.9645,
"step": 3690
},
{
"epoch": 4.99,
"learning_rate": 3.4381105632794756e-06,
"loss": 0.8626,
"step": 3700
},
{
"epoch": 5.0,
"learning_rate": 3.3493649053890326e-06,
"loss": 0.785,
"step": 3710
}
],
"logging_steps": 10,
"max_steps": 4452,
"num_train_epochs": 6,
"save_steps": 500,
"total_flos": 1.93884008030208e+16,
"trial_name": null,
"trial_params": null
}