gogpt-7b / trainer_state.json
Chriskuei's picture
init
03d29ec
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9998273778698429,
"global_step": 2896,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.993093922651934e-05,
"loss": 1.9454,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 1.9861878453038676e-05,
"loss": 1.9111,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 1.979281767955801e-05,
"loss": 1.9728,
"step": 30
},
{
"epoch": 0.01,
"learning_rate": 1.972375690607735e-05,
"loss": 1.8986,
"step": 40
},
{
"epoch": 0.02,
"learning_rate": 1.9654696132596688e-05,
"loss": 1.8296,
"step": 50
},
{
"epoch": 0.02,
"learning_rate": 1.9585635359116023e-05,
"loss": 1.8251,
"step": 60
},
{
"epoch": 0.02,
"learning_rate": 1.9516574585635362e-05,
"loss": 1.8381,
"step": 70
},
{
"epoch": 0.03,
"learning_rate": 1.9447513812154697e-05,
"loss": 1.8186,
"step": 80
},
{
"epoch": 0.03,
"learning_rate": 1.9378453038674036e-05,
"loss": 1.862,
"step": 90
},
{
"epoch": 0.03,
"learning_rate": 1.930939226519337e-05,
"loss": 1.8683,
"step": 100
},
{
"epoch": 0.04,
"learning_rate": 1.924033149171271e-05,
"loss": 1.7295,
"step": 110
},
{
"epoch": 0.04,
"learning_rate": 1.9171270718232045e-05,
"loss": 1.8422,
"step": 120
},
{
"epoch": 0.04,
"learning_rate": 1.910220994475138e-05,
"loss": 1.8391,
"step": 130
},
{
"epoch": 0.05,
"learning_rate": 1.903314917127072e-05,
"loss": 1.7898,
"step": 140
},
{
"epoch": 0.05,
"learning_rate": 1.8964088397790058e-05,
"loss": 1.8654,
"step": 150
},
{
"epoch": 0.06,
"learning_rate": 1.8895027624309393e-05,
"loss": 1.7978,
"step": 160
},
{
"epoch": 0.06,
"learning_rate": 1.882596685082873e-05,
"loss": 1.8074,
"step": 170
},
{
"epoch": 0.06,
"learning_rate": 1.8756906077348067e-05,
"loss": 1.8771,
"step": 180
},
{
"epoch": 0.07,
"learning_rate": 1.8687845303867406e-05,
"loss": 1.8324,
"step": 190
},
{
"epoch": 0.07,
"learning_rate": 1.861878453038674e-05,
"loss": 1.8138,
"step": 200
},
{
"epoch": 0.07,
"learning_rate": 1.854972375690608e-05,
"loss": 1.7218,
"step": 210
},
{
"epoch": 0.08,
"learning_rate": 1.8480662983425418e-05,
"loss": 1.7777,
"step": 220
},
{
"epoch": 0.08,
"learning_rate": 1.8411602209944753e-05,
"loss": 1.9651,
"step": 230
},
{
"epoch": 0.08,
"learning_rate": 1.834254143646409e-05,
"loss": 1.7814,
"step": 240
},
{
"epoch": 0.09,
"learning_rate": 1.8273480662983427e-05,
"loss": 1.8694,
"step": 250
},
{
"epoch": 0.09,
"learning_rate": 1.8204419889502766e-05,
"loss": 1.8094,
"step": 260
},
{
"epoch": 0.09,
"learning_rate": 1.81353591160221e-05,
"loss": 1.8144,
"step": 270
},
{
"epoch": 0.1,
"learning_rate": 1.8066298342541436e-05,
"loss": 1.7834,
"step": 280
},
{
"epoch": 0.1,
"learning_rate": 1.7997237569060775e-05,
"loss": 1.8326,
"step": 290
},
{
"epoch": 0.1,
"learning_rate": 1.7928176795580114e-05,
"loss": 1.8524,
"step": 300
},
{
"epoch": 0.11,
"learning_rate": 1.785911602209945e-05,
"loss": 1.8125,
"step": 310
},
{
"epoch": 0.11,
"learning_rate": 1.7790055248618784e-05,
"loss": 1.7685,
"step": 320
},
{
"epoch": 0.11,
"learning_rate": 1.7720994475138123e-05,
"loss": 1.8259,
"step": 330
},
{
"epoch": 0.12,
"learning_rate": 1.765193370165746e-05,
"loss": 1.8165,
"step": 340
},
{
"epoch": 0.12,
"learning_rate": 1.7582872928176797e-05,
"loss": 1.734,
"step": 350
},
{
"epoch": 0.12,
"learning_rate": 1.7513812154696135e-05,
"loss": 1.8719,
"step": 360
},
{
"epoch": 0.13,
"learning_rate": 1.744475138121547e-05,
"loss": 1.84,
"step": 370
},
{
"epoch": 0.13,
"learning_rate": 1.7375690607734806e-05,
"loss": 1.739,
"step": 380
},
{
"epoch": 0.13,
"learning_rate": 1.7306629834254145e-05,
"loss": 1.7121,
"step": 390
},
{
"epoch": 0.14,
"learning_rate": 1.7237569060773483e-05,
"loss": 1.7994,
"step": 400
},
{
"epoch": 0.14,
"learning_rate": 1.716850828729282e-05,
"loss": 1.8029,
"step": 410
},
{
"epoch": 0.15,
"learning_rate": 1.7099447513812154e-05,
"loss": 1.7876,
"step": 420
},
{
"epoch": 0.15,
"learning_rate": 1.7030386740331492e-05,
"loss": 1.7399,
"step": 430
},
{
"epoch": 0.15,
"learning_rate": 1.696132596685083e-05,
"loss": 1.8004,
"step": 440
},
{
"epoch": 0.16,
"learning_rate": 1.6892265193370166e-05,
"loss": 1.8154,
"step": 450
},
{
"epoch": 0.16,
"learning_rate": 1.6823204419889505e-05,
"loss": 1.7839,
"step": 460
},
{
"epoch": 0.16,
"learning_rate": 1.675414364640884e-05,
"loss": 1.7881,
"step": 470
},
{
"epoch": 0.17,
"learning_rate": 1.668508287292818e-05,
"loss": 1.7409,
"step": 480
},
{
"epoch": 0.17,
"learning_rate": 1.6616022099447514e-05,
"loss": 1.7907,
"step": 490
},
{
"epoch": 0.17,
"learning_rate": 1.6546961325966853e-05,
"loss": 1.8985,
"step": 500
},
{
"epoch": 0.18,
"learning_rate": 1.647790055248619e-05,
"loss": 1.8918,
"step": 510
},
{
"epoch": 0.18,
"learning_rate": 1.6408839779005527e-05,
"loss": 1.8328,
"step": 520
},
{
"epoch": 0.18,
"learning_rate": 1.6339779005524862e-05,
"loss": 1.7157,
"step": 530
},
{
"epoch": 0.19,
"learning_rate": 1.62707182320442e-05,
"loss": 1.6767,
"step": 540
},
{
"epoch": 0.19,
"learning_rate": 1.620165745856354e-05,
"loss": 1.7403,
"step": 550
},
{
"epoch": 0.19,
"learning_rate": 1.6132596685082875e-05,
"loss": 1.8535,
"step": 560
},
{
"epoch": 0.2,
"learning_rate": 1.606353591160221e-05,
"loss": 1.8406,
"step": 570
},
{
"epoch": 0.2,
"learning_rate": 1.599447513812155e-05,
"loss": 1.7262,
"step": 580
},
{
"epoch": 0.2,
"learning_rate": 1.5925414364640884e-05,
"loss": 1.7585,
"step": 590
},
{
"epoch": 0.21,
"learning_rate": 1.5856353591160222e-05,
"loss": 1.7821,
"step": 600
},
{
"epoch": 0.21,
"learning_rate": 1.578729281767956e-05,
"loss": 1.839,
"step": 610
},
{
"epoch": 0.21,
"learning_rate": 1.5718232044198896e-05,
"loss": 1.8768,
"step": 620
},
{
"epoch": 0.22,
"learning_rate": 1.564917127071823e-05,
"loss": 1.772,
"step": 630
},
{
"epoch": 0.22,
"learning_rate": 1.558011049723757e-05,
"loss": 1.7883,
"step": 640
},
{
"epoch": 0.22,
"learning_rate": 1.551104972375691e-05,
"loss": 1.6712,
"step": 650
},
{
"epoch": 0.23,
"learning_rate": 1.5441988950276244e-05,
"loss": 1.7631,
"step": 660
},
{
"epoch": 0.23,
"learning_rate": 1.537292817679558e-05,
"loss": 1.7849,
"step": 670
},
{
"epoch": 0.23,
"learning_rate": 1.5303867403314918e-05,
"loss": 1.6738,
"step": 680
},
{
"epoch": 0.24,
"learning_rate": 1.5234806629834255e-05,
"loss": 1.8382,
"step": 690
},
{
"epoch": 0.24,
"learning_rate": 1.5165745856353594e-05,
"loss": 1.7811,
"step": 700
},
{
"epoch": 0.25,
"learning_rate": 1.5096685082872929e-05,
"loss": 1.7594,
"step": 710
},
{
"epoch": 0.25,
"learning_rate": 1.5027624309392266e-05,
"loss": 1.7665,
"step": 720
},
{
"epoch": 0.25,
"learning_rate": 1.4958563535911603e-05,
"loss": 1.7435,
"step": 730
},
{
"epoch": 0.26,
"learning_rate": 1.4889502762430942e-05,
"loss": 1.7131,
"step": 740
},
{
"epoch": 0.26,
"learning_rate": 1.4820441988950279e-05,
"loss": 1.8457,
"step": 750
},
{
"epoch": 0.26,
"learning_rate": 1.4751381215469614e-05,
"loss": 1.8551,
"step": 760
},
{
"epoch": 0.27,
"learning_rate": 1.468232044198895e-05,
"loss": 1.7981,
"step": 770
},
{
"epoch": 0.27,
"learning_rate": 1.4613259668508288e-05,
"loss": 1.8244,
"step": 780
},
{
"epoch": 0.27,
"learning_rate": 1.4544198895027626e-05,
"loss": 1.8683,
"step": 790
},
{
"epoch": 0.28,
"learning_rate": 1.4475138121546963e-05,
"loss": 1.7665,
"step": 800
},
{
"epoch": 0.28,
"learning_rate": 1.4406077348066299e-05,
"loss": 1.7605,
"step": 810
},
{
"epoch": 0.28,
"learning_rate": 1.4337016574585636e-05,
"loss": 1.8339,
"step": 820
},
{
"epoch": 0.29,
"learning_rate": 1.4267955801104974e-05,
"loss": 1.8634,
"step": 830
},
{
"epoch": 0.29,
"learning_rate": 1.4198895027624311e-05,
"loss": 1.7421,
"step": 840
},
{
"epoch": 0.29,
"learning_rate": 1.4129834254143648e-05,
"loss": 1.7681,
"step": 850
},
{
"epoch": 0.3,
"learning_rate": 1.4060773480662983e-05,
"loss": 1.749,
"step": 860
},
{
"epoch": 0.3,
"learning_rate": 1.399171270718232e-05,
"loss": 1.7926,
"step": 870
},
{
"epoch": 0.3,
"learning_rate": 1.3922651933701659e-05,
"loss": 1.7425,
"step": 880
},
{
"epoch": 0.31,
"learning_rate": 1.3853591160220996e-05,
"loss": 1.7008,
"step": 890
},
{
"epoch": 0.31,
"learning_rate": 1.3784530386740333e-05,
"loss": 1.6872,
"step": 900
},
{
"epoch": 0.31,
"learning_rate": 1.3715469613259668e-05,
"loss": 1.7296,
"step": 910
},
{
"epoch": 0.32,
"learning_rate": 1.3646408839779007e-05,
"loss": 1.7473,
"step": 920
},
{
"epoch": 0.32,
"learning_rate": 1.3577348066298344e-05,
"loss": 1.7689,
"step": 930
},
{
"epoch": 0.32,
"learning_rate": 1.350828729281768e-05,
"loss": 1.7871,
"step": 940
},
{
"epoch": 0.33,
"learning_rate": 1.343922651933702e-05,
"loss": 1.7677,
"step": 950
},
{
"epoch": 0.33,
"learning_rate": 1.3370165745856355e-05,
"loss": 1.7477,
"step": 960
},
{
"epoch": 0.33,
"learning_rate": 1.3301104972375692e-05,
"loss": 1.8677,
"step": 970
},
{
"epoch": 0.34,
"learning_rate": 1.3232044198895029e-05,
"loss": 1.8348,
"step": 980
},
{
"epoch": 0.34,
"learning_rate": 1.3162983425414365e-05,
"loss": 1.7448,
"step": 990
},
{
"epoch": 0.35,
"learning_rate": 1.3093922651933704e-05,
"loss": 1.7743,
"step": 1000
},
{
"epoch": 0.35,
"learning_rate": 1.302486187845304e-05,
"loss": 1.6812,
"step": 1010
},
{
"epoch": 0.35,
"learning_rate": 1.2955801104972376e-05,
"loss": 1.7649,
"step": 1020
},
{
"epoch": 0.36,
"learning_rate": 1.2886740331491713e-05,
"loss": 1.8005,
"step": 1030
},
{
"epoch": 0.36,
"learning_rate": 1.2817679558011052e-05,
"loss": 1.8059,
"step": 1040
},
{
"epoch": 0.36,
"learning_rate": 1.2748618784530389e-05,
"loss": 1.7549,
"step": 1050
},
{
"epoch": 0.37,
"learning_rate": 1.2679558011049724e-05,
"loss": 1.7787,
"step": 1060
},
{
"epoch": 0.37,
"learning_rate": 1.2610497237569061e-05,
"loss": 1.722,
"step": 1070
},
{
"epoch": 0.37,
"learning_rate": 1.25414364640884e-05,
"loss": 1.7192,
"step": 1080
},
{
"epoch": 0.38,
"learning_rate": 1.2472375690607737e-05,
"loss": 1.8681,
"step": 1090
},
{
"epoch": 0.38,
"learning_rate": 1.2403314917127074e-05,
"loss": 1.8199,
"step": 1100
},
{
"epoch": 0.38,
"learning_rate": 1.2334254143646409e-05,
"loss": 1.8003,
"step": 1110
},
{
"epoch": 0.39,
"learning_rate": 1.2265193370165746e-05,
"loss": 1.8412,
"step": 1120
},
{
"epoch": 0.39,
"learning_rate": 1.2196132596685085e-05,
"loss": 1.7463,
"step": 1130
},
{
"epoch": 0.39,
"learning_rate": 1.2127071823204422e-05,
"loss": 1.8106,
"step": 1140
},
{
"epoch": 0.4,
"learning_rate": 1.2058011049723757e-05,
"loss": 1.6872,
"step": 1150
},
{
"epoch": 0.4,
"learning_rate": 1.1988950276243094e-05,
"loss": 1.6963,
"step": 1160
},
{
"epoch": 0.4,
"learning_rate": 1.1919889502762432e-05,
"loss": 1.7752,
"step": 1170
},
{
"epoch": 0.41,
"learning_rate": 1.185082872928177e-05,
"loss": 1.8593,
"step": 1180
},
{
"epoch": 0.41,
"learning_rate": 1.1781767955801106e-05,
"loss": 1.7515,
"step": 1190
},
{
"epoch": 0.41,
"learning_rate": 1.1712707182320442e-05,
"loss": 1.7299,
"step": 1200
},
{
"epoch": 0.42,
"learning_rate": 1.1643646408839779e-05,
"loss": 1.7632,
"step": 1210
},
{
"epoch": 0.42,
"learning_rate": 1.1574585635359117e-05,
"loss": 1.7982,
"step": 1220
},
{
"epoch": 0.42,
"learning_rate": 1.1505524861878454e-05,
"loss": 1.7833,
"step": 1230
},
{
"epoch": 0.43,
"learning_rate": 1.1436464088397791e-05,
"loss": 1.7549,
"step": 1240
},
{
"epoch": 0.43,
"learning_rate": 1.1367403314917126e-05,
"loss": 1.726,
"step": 1250
},
{
"epoch": 0.44,
"learning_rate": 1.1298342541436465e-05,
"loss": 1.7221,
"step": 1260
},
{
"epoch": 0.44,
"learning_rate": 1.1229281767955802e-05,
"loss": 1.8503,
"step": 1270
},
{
"epoch": 0.44,
"learning_rate": 1.1160220994475139e-05,
"loss": 1.7096,
"step": 1280
},
{
"epoch": 0.45,
"learning_rate": 1.1091160220994478e-05,
"loss": 1.7261,
"step": 1290
},
{
"epoch": 0.45,
"learning_rate": 1.1022099447513813e-05,
"loss": 1.7823,
"step": 1300
},
{
"epoch": 0.45,
"learning_rate": 1.095303867403315e-05,
"loss": 1.7031,
"step": 1310
},
{
"epoch": 0.46,
"learning_rate": 1.0883977900552487e-05,
"loss": 1.7445,
"step": 1320
},
{
"epoch": 0.46,
"learning_rate": 1.0814917127071824e-05,
"loss": 1.7109,
"step": 1330
},
{
"epoch": 0.46,
"learning_rate": 1.0745856353591162e-05,
"loss": 1.8005,
"step": 1340
},
{
"epoch": 0.47,
"learning_rate": 1.0676795580110498e-05,
"loss": 1.7529,
"step": 1350
},
{
"epoch": 0.47,
"learning_rate": 1.0607734806629835e-05,
"loss": 1.6576,
"step": 1360
},
{
"epoch": 0.47,
"learning_rate": 1.0538674033149172e-05,
"loss": 1.7873,
"step": 1370
},
{
"epoch": 0.48,
"learning_rate": 1.046961325966851e-05,
"loss": 1.7714,
"step": 1380
},
{
"epoch": 0.48,
"learning_rate": 1.0400552486187847e-05,
"loss": 1.7174,
"step": 1390
},
{
"epoch": 0.48,
"learning_rate": 1.0331491712707182e-05,
"loss": 1.6726,
"step": 1400
},
{
"epoch": 0.49,
"learning_rate": 1.026243093922652e-05,
"loss": 1.718,
"step": 1410
},
{
"epoch": 0.49,
"learning_rate": 1.0193370165745858e-05,
"loss": 1.7388,
"step": 1420
},
{
"epoch": 0.49,
"learning_rate": 1.0124309392265195e-05,
"loss": 1.7361,
"step": 1430
},
{
"epoch": 0.5,
"learning_rate": 1.0055248618784532e-05,
"loss": 1.7668,
"step": 1440
},
{
"epoch": 0.5,
"learning_rate": 9.986187845303869e-06,
"loss": 1.8001,
"step": 1450
},
{
"epoch": 0.5,
"learning_rate": 9.917127071823204e-06,
"loss": 1.7302,
"step": 1460
},
{
"epoch": 0.51,
"learning_rate": 9.848066298342543e-06,
"loss": 1.687,
"step": 1470
},
{
"epoch": 0.51,
"learning_rate": 9.779005524861878e-06,
"loss": 1.6978,
"step": 1480
},
{
"epoch": 0.51,
"learning_rate": 9.709944751381217e-06,
"loss": 1.7913,
"step": 1490
},
{
"epoch": 0.52,
"learning_rate": 9.640883977900554e-06,
"loss": 1.8389,
"step": 1500
},
{
"epoch": 0.52,
"learning_rate": 9.57182320441989e-06,
"loss": 1.8108,
"step": 1510
},
{
"epoch": 0.52,
"learning_rate": 9.502762430939228e-06,
"loss": 1.7494,
"step": 1520
},
{
"epoch": 0.53,
"learning_rate": 9.433701657458565e-06,
"loss": 1.8438,
"step": 1530
},
{
"epoch": 0.53,
"learning_rate": 9.364640883977902e-06,
"loss": 1.8194,
"step": 1540
},
{
"epoch": 0.54,
"learning_rate": 9.295580110497238e-06,
"loss": 1.7495,
"step": 1550
},
{
"epoch": 0.54,
"learning_rate": 9.226519337016575e-06,
"loss": 1.7208,
"step": 1560
},
{
"epoch": 0.54,
"learning_rate": 9.157458563535912e-06,
"loss": 1.706,
"step": 1570
},
{
"epoch": 0.55,
"learning_rate": 9.08839779005525e-06,
"loss": 1.8367,
"step": 1580
},
{
"epoch": 0.55,
"learning_rate": 9.019337016574586e-06,
"loss": 1.6874,
"step": 1590
},
{
"epoch": 0.55,
"learning_rate": 8.950276243093923e-06,
"loss": 1.7478,
"step": 1600
},
{
"epoch": 0.56,
"learning_rate": 8.88121546961326e-06,
"loss": 1.6782,
"step": 1610
},
{
"epoch": 0.56,
"learning_rate": 8.812154696132597e-06,
"loss": 1.8467,
"step": 1620
},
{
"epoch": 0.56,
"learning_rate": 8.743093922651934e-06,
"loss": 1.758,
"step": 1630
},
{
"epoch": 0.57,
"learning_rate": 8.674033149171271e-06,
"loss": 1.7116,
"step": 1640
},
{
"epoch": 0.57,
"learning_rate": 8.60497237569061e-06,
"loss": 1.751,
"step": 1650
},
{
"epoch": 0.57,
"learning_rate": 8.535911602209945e-06,
"loss": 1.7502,
"step": 1660
},
{
"epoch": 0.58,
"learning_rate": 8.466850828729282e-06,
"loss": 1.879,
"step": 1670
},
{
"epoch": 0.58,
"learning_rate": 8.397790055248619e-06,
"loss": 1.8171,
"step": 1680
},
{
"epoch": 0.58,
"learning_rate": 8.328729281767956e-06,
"loss": 1.7586,
"step": 1690
},
{
"epoch": 0.59,
"learning_rate": 8.259668508287293e-06,
"loss": 1.7848,
"step": 1700
},
{
"epoch": 0.59,
"learning_rate": 8.19060773480663e-06,
"loss": 1.7789,
"step": 1710
},
{
"epoch": 0.59,
"learning_rate": 8.121546961325968e-06,
"loss": 1.7104,
"step": 1720
},
{
"epoch": 0.6,
"learning_rate": 8.052486187845304e-06,
"loss": 1.807,
"step": 1730
},
{
"epoch": 0.6,
"learning_rate": 7.983425414364642e-06,
"loss": 1.6772,
"step": 1740
},
{
"epoch": 0.6,
"learning_rate": 7.914364640883978e-06,
"loss": 1.7574,
"step": 1750
},
{
"epoch": 0.61,
"learning_rate": 7.845303867403316e-06,
"loss": 1.7925,
"step": 1760
},
{
"epoch": 0.61,
"learning_rate": 7.776243093922653e-06,
"loss": 1.7109,
"step": 1770
},
{
"epoch": 0.61,
"learning_rate": 7.70718232044199e-06,
"loss": 1.8161,
"step": 1780
},
{
"epoch": 0.62,
"learning_rate": 7.638121546961327e-06,
"loss": 1.716,
"step": 1790
},
{
"epoch": 0.62,
"learning_rate": 7.569060773480663e-06,
"loss": 1.7228,
"step": 1800
},
{
"epoch": 0.62,
"learning_rate": 7.500000000000001e-06,
"loss": 1.865,
"step": 1810
},
{
"epoch": 0.63,
"learning_rate": 7.430939226519338e-06,
"loss": 1.7947,
"step": 1820
},
{
"epoch": 0.63,
"learning_rate": 7.361878453038674e-06,
"loss": 1.8093,
"step": 1830
},
{
"epoch": 0.64,
"learning_rate": 7.292817679558012e-06,
"loss": 1.7586,
"step": 1840
},
{
"epoch": 0.64,
"learning_rate": 7.223756906077348e-06,
"loss": 1.7256,
"step": 1850
},
{
"epoch": 0.64,
"learning_rate": 7.154696132596686e-06,
"loss": 1.7485,
"step": 1860
},
{
"epoch": 0.65,
"learning_rate": 7.085635359116023e-06,
"loss": 1.7597,
"step": 1870
},
{
"epoch": 0.65,
"learning_rate": 7.01657458563536e-06,
"loss": 1.7516,
"step": 1880
},
{
"epoch": 0.65,
"learning_rate": 6.947513812154697e-06,
"loss": 1.7449,
"step": 1890
},
{
"epoch": 0.66,
"learning_rate": 6.878453038674034e-06,
"loss": 1.8367,
"step": 1900
},
{
"epoch": 0.66,
"learning_rate": 6.809392265193371e-06,
"loss": 1.7401,
"step": 1910
},
{
"epoch": 0.66,
"learning_rate": 6.740331491712708e-06,
"loss": 1.8057,
"step": 1920
},
{
"epoch": 0.67,
"learning_rate": 6.6712707182320445e-06,
"loss": 1.7072,
"step": 1930
},
{
"epoch": 0.67,
"learning_rate": 6.602209944751382e-06,
"loss": 1.7182,
"step": 1940
},
{
"epoch": 0.67,
"learning_rate": 6.5331491712707184e-06,
"loss": 1.7331,
"step": 1950
},
{
"epoch": 0.68,
"learning_rate": 6.464088397790056e-06,
"loss": 1.871,
"step": 1960
},
{
"epoch": 0.68,
"learning_rate": 6.395027624309392e-06,
"loss": 1.7339,
"step": 1970
},
{
"epoch": 0.68,
"learning_rate": 6.32596685082873e-06,
"loss": 1.7293,
"step": 1980
},
{
"epoch": 0.69,
"learning_rate": 6.256906077348067e-06,
"loss": 1.8179,
"step": 1990
},
{
"epoch": 0.69,
"learning_rate": 6.187845303867403e-06,
"loss": 1.7189,
"step": 2000
},
{
"epoch": 0.69,
"learning_rate": 6.118784530386741e-06,
"loss": 1.6855,
"step": 2010
},
{
"epoch": 0.7,
"learning_rate": 6.049723756906077e-06,
"loss": 1.8562,
"step": 2020
},
{
"epoch": 0.7,
"learning_rate": 5.980662983425415e-06,
"loss": 1.7805,
"step": 2030
},
{
"epoch": 0.7,
"learning_rate": 5.911602209944752e-06,
"loss": 1.7376,
"step": 2040
},
{
"epoch": 0.71,
"learning_rate": 5.842541436464089e-06,
"loss": 1.7232,
"step": 2050
},
{
"epoch": 0.71,
"learning_rate": 5.773480662983426e-06,
"loss": 1.7372,
"step": 2060
},
{
"epoch": 0.71,
"learning_rate": 5.704419889502763e-06,
"loss": 1.6427,
"step": 2070
},
{
"epoch": 0.72,
"learning_rate": 5.6353591160221e-06,
"loss": 1.6946,
"step": 2080
},
{
"epoch": 0.72,
"learning_rate": 5.5662983425414375e-06,
"loss": 1.7831,
"step": 2090
},
{
"epoch": 0.73,
"learning_rate": 5.497237569060774e-06,
"loss": 1.8121,
"step": 2100
},
{
"epoch": 0.73,
"learning_rate": 5.4281767955801114e-06,
"loss": 1.7363,
"step": 2110
},
{
"epoch": 0.73,
"learning_rate": 5.3591160220994476e-06,
"loss": 1.7434,
"step": 2120
},
{
"epoch": 0.74,
"learning_rate": 5.290055248618785e-06,
"loss": 1.8479,
"step": 2130
},
{
"epoch": 0.74,
"learning_rate": 5.220994475138122e-06,
"loss": 1.7347,
"step": 2140
},
{
"epoch": 0.74,
"learning_rate": 5.151933701657459e-06,
"loss": 1.785,
"step": 2150
},
{
"epoch": 0.75,
"learning_rate": 5.082872928176796e-06,
"loss": 1.8255,
"step": 2160
},
{
"epoch": 0.75,
"learning_rate": 5.013812154696132e-06,
"loss": 1.7684,
"step": 2170
},
{
"epoch": 0.75,
"learning_rate": 4.94475138121547e-06,
"loss": 1.7494,
"step": 2180
},
{
"epoch": 0.76,
"learning_rate": 4.875690607734807e-06,
"loss": 1.6692,
"step": 2190
},
{
"epoch": 0.76,
"learning_rate": 4.806629834254144e-06,
"loss": 1.7653,
"step": 2200
},
{
"epoch": 0.76,
"learning_rate": 4.737569060773481e-06,
"loss": 1.6953,
"step": 2210
},
{
"epoch": 0.77,
"learning_rate": 4.668508287292818e-06,
"loss": 1.8247,
"step": 2220
},
{
"epoch": 0.77,
"learning_rate": 4.599447513812155e-06,
"loss": 1.8188,
"step": 2230
},
{
"epoch": 0.77,
"learning_rate": 4.530386740331492e-06,
"loss": 1.7473,
"step": 2240
},
{
"epoch": 0.78,
"learning_rate": 4.461325966850829e-06,
"loss": 1.6097,
"step": 2250
},
{
"epoch": 0.78,
"learning_rate": 4.392265193370166e-06,
"loss": 1.7937,
"step": 2260
},
{
"epoch": 0.78,
"learning_rate": 4.323204419889503e-06,
"loss": 1.7121,
"step": 2270
},
{
"epoch": 0.79,
"learning_rate": 4.2541436464088406e-06,
"loss": 1.8483,
"step": 2280
},
{
"epoch": 0.79,
"learning_rate": 4.1850828729281775e-06,
"loss": 1.7838,
"step": 2290
},
{
"epoch": 0.79,
"learning_rate": 4.1160220994475145e-06,
"loss": 1.6945,
"step": 2300
},
{
"epoch": 0.8,
"learning_rate": 4.0469613259668514e-06,
"loss": 1.7295,
"step": 2310
},
{
"epoch": 0.8,
"learning_rate": 3.977900552486188e-06,
"loss": 1.7116,
"step": 2320
},
{
"epoch": 0.8,
"learning_rate": 3.9088397790055245e-06,
"loss": 1.8388,
"step": 2330
},
{
"epoch": 0.81,
"learning_rate": 3.839779005524862e-06,
"loss": 1.8654,
"step": 2340
},
{
"epoch": 0.81,
"learning_rate": 3.7707182320441993e-06,
"loss": 1.7835,
"step": 2350
},
{
"epoch": 0.81,
"learning_rate": 3.7016574585635362e-06,
"loss": 1.6937,
"step": 2360
},
{
"epoch": 0.82,
"learning_rate": 3.632596685082873e-06,
"loss": 1.6986,
"step": 2370
},
{
"epoch": 0.82,
"learning_rate": 3.56353591160221e-06,
"loss": 1.7051,
"step": 2380
},
{
"epoch": 0.83,
"learning_rate": 3.4944751381215475e-06,
"loss": 1.7332,
"step": 2390
},
{
"epoch": 0.83,
"learning_rate": 3.4254143646408845e-06,
"loss": 1.6828,
"step": 2400
},
{
"epoch": 0.83,
"learning_rate": 3.3563535911602214e-06,
"loss": 1.818,
"step": 2410
},
{
"epoch": 0.84,
"learning_rate": 3.2872928176795584e-06,
"loss": 1.7628,
"step": 2420
},
{
"epoch": 0.84,
"learning_rate": 3.218232044198895e-06,
"loss": 1.7445,
"step": 2430
},
{
"epoch": 0.84,
"learning_rate": 3.149171270718232e-06,
"loss": 1.8389,
"step": 2440
},
{
"epoch": 0.85,
"learning_rate": 3.0801104972375697e-06,
"loss": 1.7749,
"step": 2450
},
{
"epoch": 0.85,
"learning_rate": 3.0110497237569062e-06,
"loss": 1.8421,
"step": 2460
},
{
"epoch": 0.85,
"learning_rate": 2.941988950276243e-06,
"loss": 1.7727,
"step": 2470
},
{
"epoch": 0.86,
"learning_rate": 2.87292817679558e-06,
"loss": 1.754,
"step": 2480
},
{
"epoch": 0.86,
"learning_rate": 2.803867403314917e-06,
"loss": 1.6552,
"step": 2490
},
{
"epoch": 0.86,
"learning_rate": 2.7348066298342545e-06,
"loss": 1.7354,
"step": 2500
},
{
"epoch": 0.87,
"learning_rate": 2.6657458563535914e-06,
"loss": 1.7694,
"step": 2510
},
{
"epoch": 0.87,
"learning_rate": 2.5966850828729284e-06,
"loss": 1.7042,
"step": 2520
},
{
"epoch": 0.87,
"learning_rate": 2.5276243093922653e-06,
"loss": 1.6969,
"step": 2530
},
{
"epoch": 0.88,
"learning_rate": 2.4585635359116027e-06,
"loss": 1.7404,
"step": 2540
},
{
"epoch": 0.88,
"learning_rate": 2.3895027624309393e-06,
"loss": 1.7236,
"step": 2550
},
{
"epoch": 0.88,
"learning_rate": 2.320441988950276e-06,
"loss": 1.8104,
"step": 2560
},
{
"epoch": 0.89,
"learning_rate": 2.2513812154696136e-06,
"loss": 1.7518,
"step": 2570
},
{
"epoch": 0.89,
"learning_rate": 2.1823204419889505e-06,
"loss": 1.7097,
"step": 2580
},
{
"epoch": 0.89,
"learning_rate": 2.1132596685082875e-06,
"loss": 1.7118,
"step": 2590
},
{
"epoch": 0.9,
"learning_rate": 2.0441988950276245e-06,
"loss": 1.7996,
"step": 2600
},
{
"epoch": 0.9,
"learning_rate": 1.9751381215469614e-06,
"loss": 1.6647,
"step": 2610
},
{
"epoch": 0.9,
"learning_rate": 1.9060773480662986e-06,
"loss": 1.6974,
"step": 2620
},
{
"epoch": 0.91,
"learning_rate": 1.8370165745856355e-06,
"loss": 1.6929,
"step": 2630
},
{
"epoch": 0.91,
"learning_rate": 1.7679558011049725e-06,
"loss": 1.7994,
"step": 2640
},
{
"epoch": 0.91,
"learning_rate": 1.6988950276243097e-06,
"loss": 1.7755,
"step": 2650
},
{
"epoch": 0.92,
"learning_rate": 1.6298342541436466e-06,
"loss": 1.8172,
"step": 2660
},
{
"epoch": 0.92,
"learning_rate": 1.5607734806629834e-06,
"loss": 1.8241,
"step": 2670
},
{
"epoch": 0.93,
"learning_rate": 1.4917127071823205e-06,
"loss": 1.7558,
"step": 2680
},
{
"epoch": 0.93,
"learning_rate": 1.4226519337016575e-06,
"loss": 1.7474,
"step": 2690
},
{
"epoch": 0.93,
"learning_rate": 1.3535911602209945e-06,
"loss": 1.8138,
"step": 2700
},
{
"epoch": 0.94,
"learning_rate": 1.2845303867403316e-06,
"loss": 1.7158,
"step": 2710
},
{
"epoch": 0.94,
"learning_rate": 1.2154696132596686e-06,
"loss": 1.8223,
"step": 2720
},
{
"epoch": 0.94,
"learning_rate": 1.1464088397790055e-06,
"loss": 1.8411,
"step": 2730
},
{
"epoch": 0.95,
"learning_rate": 1.0773480662983427e-06,
"loss": 1.8054,
"step": 2740
},
{
"epoch": 0.95,
"learning_rate": 1.0082872928176797e-06,
"loss": 1.7401,
"step": 2750
},
{
"epoch": 0.95,
"learning_rate": 9.392265193370166e-07,
"loss": 1.6664,
"step": 2760
},
{
"epoch": 0.96,
"learning_rate": 8.701657458563537e-07,
"loss": 1.7788,
"step": 2770
},
{
"epoch": 0.96,
"learning_rate": 8.011049723756907e-07,
"loss": 1.7135,
"step": 2780
},
{
"epoch": 0.96,
"learning_rate": 7.320441988950276e-07,
"loss": 1.8247,
"step": 2790
},
{
"epoch": 0.97,
"learning_rate": 6.629834254143647e-07,
"loss": 1.7436,
"step": 2800
},
{
"epoch": 0.97,
"learning_rate": 5.939226519337017e-07,
"loss": 1.792,
"step": 2810
},
{
"epoch": 0.97,
"learning_rate": 5.248618784530387e-07,
"loss": 1.7798,
"step": 2820
},
{
"epoch": 0.98,
"learning_rate": 4.5580110497237574e-07,
"loss": 1.7821,
"step": 2830
},
{
"epoch": 0.98,
"learning_rate": 3.867403314917127e-07,
"loss": 1.7994,
"step": 2840
},
{
"epoch": 0.98,
"learning_rate": 3.1767955801104976e-07,
"loss": 1.8054,
"step": 2850
},
{
"epoch": 0.99,
"learning_rate": 2.486187845303868e-07,
"loss": 1.7472,
"step": 2860
},
{
"epoch": 0.99,
"learning_rate": 1.7955801104972376e-07,
"loss": 1.7099,
"step": 2870
},
{
"epoch": 0.99,
"learning_rate": 1.1049723756906078e-07,
"loss": 1.7215,
"step": 2880
},
{
"epoch": 1.0,
"learning_rate": 4.143646408839779e-08,
"loss": 1.7854,
"step": 2890
},
{
"epoch": 1.0,
"step": 2896,
"total_flos": 1.202498608338174e+18,
"train_loss": 1.7728749915381163,
"train_runtime": 8665.6227,
"train_samples_per_second": 10.696,
"train_steps_per_second": 0.334
}
],
"max_steps": 2896,
"num_train_epochs": 1,
"total_flos": 1.202498608338174e+18,
"trial_name": null,
"trial_params": null
}