finetuned_models/my_bloomz_model/trainer_state.json

980 lines
19 KiB
JSON

{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.99249530956848,
"global_step": 1596,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03,
"learning_rate": 9.999031365446931e-05,
"loss": 2.7532,
"step": 10
},
{
"epoch": 0.05,
"learning_rate": 9.996125837088884e-05,
"loss": 2.5659,
"step": 20
},
{
"epoch": 0.08,
"learning_rate": 9.991284540683922e-05,
"loss": 2.4466,
"step": 30
},
{
"epoch": 0.1,
"learning_rate": 9.984509352010838e-05,
"loss": 2.3335,
"step": 40
},
{
"epoch": 0.13,
"learning_rate": 9.975802896142372e-05,
"loss": 2.34,
"step": 50
},
{
"epoch": 0.15,
"learning_rate": 9.965168546428121e-05,
"loss": 2.3421,
"step": 60
},
{
"epoch": 0.18,
"learning_rate": 9.952610423187516e-05,
"loss": 2.3537,
"step": 70
},
{
"epoch": 0.2,
"learning_rate": 9.938133392113398e-05,
"loss": 2.3439,
"step": 80
},
{
"epoch": 0.23,
"learning_rate": 9.921743062386772e-05,
"loss": 2.2641,
"step": 90
},
{
"epoch": 0.25,
"learning_rate": 9.903445784503524e-05,
"loss": 2.3106,
"step": 100
},
{
"epoch": 0.28,
"learning_rate": 9.883248647813889e-05,
"loss": 2.2766,
"step": 110
},
{
"epoch": 0.3,
"learning_rate": 9.861159477775653e-05,
"loss": 2.2866,
"step": 120
},
{
"epoch": 0.33,
"learning_rate": 9.837186832922157e-05,
"loss": 2.2493,
"step": 130
},
{
"epoch": 0.35,
"learning_rate": 9.811340001546251e-05,
"loss": 2.263,
"step": 140
},
{
"epoch": 0.38,
"learning_rate": 9.783628998101524e-05,
"loss": 2.2712,
"step": 150
},
{
"epoch": 0.4,
"learning_rate": 9.754064559322146e-05,
"loss": 2.2659,
"step": 160
},
{
"epoch": 0.43,
"learning_rate": 9.722658140062897e-05,
"loss": 2.2891,
"step": 170
},
{
"epoch": 0.45,
"learning_rate": 9.689421908860927e-05,
"loss": 2.2305,
"step": 180
},
{
"epoch": 0.48,
"learning_rate": 9.654368743221022e-05,
"loss": 2.2444,
"step": 190
},
{
"epoch": 0.5,
"learning_rate": 9.617512224626153e-05,
"loss": 2.211,
"step": 200
},
{
"epoch": 0.53,
"learning_rate": 9.578866633275288e-05,
"loss": 2.2644,
"step": 210
},
{
"epoch": 0.55,
"learning_rate": 9.538446942550469e-05,
"loss": 2.2671,
"step": 220
},
{
"epoch": 0.58,
"learning_rate": 9.496268813215319e-05,
"loss": 2.2883,
"step": 230
},
{
"epoch": 0.6,
"learning_rate": 9.452348587347224e-05,
"loss": 2.2465,
"step": 240
},
{
"epoch": 0.63,
"learning_rate": 9.406703282005522e-05,
"loss": 2.2099,
"step": 250
},
{
"epoch": 0.65,
"learning_rate": 9.359350582638192e-05,
"loss": 2.2062,
"step": 260
},
{
"epoch": 0.68,
"learning_rate": 9.310308836229549e-05,
"loss": 2.2527,
"step": 270
},
{
"epoch": 0.7,
"learning_rate": 9.259597044191636e-05,
"loss": 2.271,
"step": 280
},
{
"epoch": 0.73,
"learning_rate": 9.207234855002063e-05,
"loss": 2.2306,
"step": 290
},
{
"epoch": 0.75,
"learning_rate": 9.153242556591116e-05,
"loss": 2.1917,
"step": 300
},
{
"epoch": 0.78,
"learning_rate": 9.097641068481133e-05,
"loss": 2.1985,
"step": 310
},
{
"epoch": 0.8,
"learning_rate": 9.040451933681148e-05,
"loss": 2.2027,
"step": 320
},
{
"epoch": 0.83,
"learning_rate": 8.981697310339972e-05,
"loss": 2.2017,
"step": 330
},
{
"epoch": 0.85,
"learning_rate": 8.921399963160933e-05,
"loss": 2.2355,
"step": 340
},
{
"epoch": 0.88,
"learning_rate": 8.859583254581605e-05,
"loss": 2.2395,
"step": 350
},
{
"epoch": 0.9,
"learning_rate": 8.796271135721944e-05,
"loss": 2.2057,
"step": 360
},
{
"epoch": 0.93,
"learning_rate": 8.731488137104331e-05,
"loss": 2.2407,
"step": 370
},
{
"epoch": 0.95,
"learning_rate": 8.665259359149132e-05,
"loss": 2.2288,
"step": 380
},
{
"epoch": 0.98,
"learning_rate": 8.597610462449442e-05,
"loss": 2.2283,
"step": 390
},
{
"epoch": 1.0,
"learning_rate": 8.528567657828784e-05,
"loss": 2.2171,
"step": 400
},
{
"epoch": 1.03,
"learning_rate": 8.4652594410225e-05,
"loss": 2.213,
"step": 410
},
{
"epoch": 1.05,
"learning_rate": 8.400863688854597e-05,
"loss": 2.1943,
"step": 420
},
{
"epoch": 1.08,
"learning_rate": 8.32806194092743e-05,
"loss": 2.1694,
"step": 430
},
{
"epoch": 1.1,
"learning_rate": 8.253970722683968e-05,
"loss": 2.196,
"step": 440
},
{
"epoch": 1.13,
"learning_rate": 8.178618741049842e-05,
"loss": 2.2067,
"step": 450
},
{
"epoch": 1.15,
"learning_rate": 8.102035191438266e-05,
"loss": 2.2222,
"step": 460
},
{
"epoch": 1.18,
"learning_rate": 8.024249746438188e-05,
"loss": 2.1976,
"step": 470
},
{
"epoch": 1.2,
"learning_rate": 7.945292544317505e-05,
"loss": 2.197,
"step": 480
},
{
"epoch": 1.23,
"learning_rate": 7.865194177345893e-05,
"loss": 2.1796,
"step": 490
},
{
"epoch": 1.25,
"learning_rate": 7.78398567994171e-05,
"loss": 2.1592,
"step": 500
},
{
"epoch": 1.28,
"learning_rate": 7.70169851664759e-05,
"loss": 2.1824,
"step": 510
},
{
"epoch": 1.3,
"learning_rate": 7.618364569939391e-05,
"loss": 2.1466,
"step": 520
},
{
"epoch": 1.33,
"learning_rate": 7.550965412779563e-05,
"loss": 2.2008,
"step": 530
},
{
"epoch": 1.35,
"learning_rate": 7.465828886394729e-05,
"loss": 2.1515,
"step": 540
},
{
"epoch": 1.38,
"learning_rate": 7.379736965185368e-05,
"loss": 2.1821,
"step": 550
},
{
"epoch": 1.4,
"learning_rate": 7.29272300579533e-05,
"loss": 2.144,
"step": 560
},
{
"epoch": 1.43,
"learning_rate": 7.21364995705929e-05,
"loss": 2.1473,
"step": 570
},
{
"epoch": 1.45,
"learning_rate": 7.124977291636525e-05,
"loss": 2.1695,
"step": 580
},
{
"epoch": 1.48,
"learning_rate": 7.035481295642096e-05,
"loss": 2.2053,
"step": 590
},
{
"epoch": 1.5,
"learning_rate": 6.94519664464163e-05,
"loss": 2.2068,
"step": 600
},
{
"epoch": 1.53,
"learning_rate": 6.85415831976816e-05,
"loss": 2.1685,
"step": 610
},
{
"epoch": 1.55,
"learning_rate": 6.762401594168536e-05,
"loss": 2.198,
"step": 620
},
{
"epoch": 1.58,
"learning_rate": 6.669962019336713e-05,
"loss": 2.1921,
"step": 630
},
{
"epoch": 1.6,
"learning_rate": 6.576875411339194e-05,
"loss": 2.2161,
"step": 640
},
{
"epoch": 1.63,
"learning_rate": 6.483177836937949e-05,
"loss": 2.1481,
"step": 650
},
{
"epoch": 1.65,
"learning_rate": 6.388905599616225e-05,
"loss": 2.1845,
"step": 660
},
{
"epoch": 1.68,
"learning_rate": 6.294095225512603e-05,
"loss": 2.194,
"step": 670
},
{
"epoch": 1.7,
"learning_rate": 6.198783449268826e-05,
"loss": 2.1456,
"step": 680
},
{
"epoch": 1.73,
"learning_rate": 6.112604669781572e-05,
"loss": 2.213,
"step": 690
},
{
"epoch": 1.75,
"learning_rate": 6.016442117659135e-05,
"loss": 2.1536,
"step": 700
},
{
"epoch": 1.78,
"learning_rate": 5.9198857411541544e-05,
"loss": 2.1964,
"step": 710
},
{
"epoch": 1.8,
"learning_rate": 5.8229729514036705e-05,
"loss": 2.1609,
"step": 720
},
{
"epoch": 1.83,
"learning_rate": 5.7257412976383983e-05,
"loss": 2.1863,
"step": 730
},
{
"epoch": 1.85,
"learning_rate": 5.6282284526341324e-05,
"loss": 2.1919,
"step": 740
},
{
"epoch": 1.88,
"learning_rate": 5.5304721981152906e-05,
"loss": 2.1481,
"step": 750
},
{
"epoch": 1.9,
"learning_rate": 5.432510410116234e-05,
"loss": 2.1514,
"step": 760
},
{
"epoch": 1.93,
"learning_rate": 5.334381044306057e-05,
"loss": 2.1453,
"step": 770
},
{
"epoch": 1.95,
"learning_rate": 5.236122121282517e-05,
"loss": 2.205,
"step": 780
},
{
"epoch": 1.98,
"learning_rate": 5.157447283934396e-05,
"loss": 2.1836,
"step": 790
},
{
"epoch": 2.0,
"learning_rate": 5.059051120646924e-05,
"loss": 2.1963,
"step": 800
},
{
"epoch": 2.03,
"learning_rate": 4.9606320777771064e-05,
"loss": 2.164,
"step": 810
},
{
"epoch": 2.05,
"learning_rate": 4.8622282881591906e-05,
"loss": 2.1199,
"step": 820
},
{
"epoch": 2.08,
"learning_rate": 4.763877878717484e-05,
"loss": 2.1145,
"step": 830
},
{
"epoch": 2.1,
"learning_rate": 4.665618955693944e-05,
"loss": 2.1292,
"step": 840
},
{
"epoch": 2.13,
"learning_rate": 4.5674895898837664e-05,
"loss": 2.111,
"step": 850
},
{
"epoch": 2.15,
"learning_rate": 4.4695278018847105e-05,
"loss": 2.1586,
"step": 860
},
{
"epoch": 2.18,
"learning_rate": 4.371771547365869e-05,
"loss": 2.1526,
"step": 870
},
{
"epoch": 2.2,
"learning_rate": 4.2742587023616035e-05,
"loss": 2.1504,
"step": 880
},
{
"epoch": 2.23,
"learning_rate": 4.196449074070329e-05,
"loss": 2.1642,
"step": 890
},
{
"epoch": 2.25,
"learning_rate": 4.109152378329036e-05,
"loss": 2.0918,
"step": 900
},
{
"epoch": 2.28,
"learning_rate": 4.012485143116954e-05,
"loss": 2.1696,
"step": 910
},
{
"epoch": 2.3,
"learning_rate": 3.916200524309692e-05,
"loss": 2.1041,
"step": 920
},
{
"epoch": 2.33,
"learning_rate": 3.820335827750731e-05,
"loss": 2.1478,
"step": 930
},
{
"epoch": 2.35,
"learning_rate": 3.724928196583073e-05,
"loss": 2.121,
"step": 940
},
{
"epoch": 2.38,
"learning_rate": 3.6300145968579876e-05,
"loss": 2.153,
"step": 950
},
{
"epoch": 2.4,
"learning_rate": 3.535631803212376e-05,
"loss": 2.1578,
"step": 960
},
{
"epoch": 2.43,
"learning_rate": 3.4418163846202944e-05,
"loss": 2.0972,
"step": 970
},
{
"epoch": 2.45,
"learning_rate": 3.3486046902241664e-05,
"loss": 2.1384,
"step": 980
},
{
"epoch": 2.48,
"learning_rate": 3.256032835251166e-05,
"loss": 2.1403,
"step": 990
},
{
"epoch": 2.5,
"learning_rate": 3.164136687020244e-05,
"loss": 2.1418,
"step": 1000
},
{
"epoch": 2.53,
"learning_rate": 3.082037316466236e-05,
"loss": 2.123,
"step": 1010
},
{
"epoch": 2.55,
"learning_rate": 2.991522876735154e-05,
"loss": 2.1421,
"step": 1020
},
{
"epoch": 2.58,
"learning_rate": 2.9017866291403273e-05,
"loss": 2.1053,
"step": 1030
},
{
"epoch": 2.6,
"learning_rate": 2.812863342333793e-05,
"loss": 2.1515,
"step": 1040
},
{
"epoch": 2.63,
"learning_rate": 2.7247874699828185e-05,
"loss": 2.1383,
"step": 1050
},
{
"epoch": 2.65,
"learning_rate": 2.637593137420702e-05,
"loss": 2.1525,
"step": 1060
},
{
"epoch": 2.68,
"learning_rate": 2.5513141284247876e-05,
"loss": 2.1938,
"step": 1070
},
{
"epoch": 2.7,
"learning_rate": 2.474473258691967e-05,
"loss": 2.0892,
"step": 1080
},
{
"epoch": 2.73,
"learning_rate": 2.390025158175458e-05,
"loss": 2.1276,
"step": 1090
},
{
"epoch": 2.75,
"learning_rate": 2.306588302384722e-05,
"loss": 2.1428,
"step": 1100
},
{
"epoch": 2.78,
"learning_rate": 2.2241950192483658e-05,
"loss": 2.1415,
"step": 1110
},
{
"epoch": 2.8,
"learning_rate": 2.1428772323587827e-05,
"loss": 2.1729,
"step": 1120
},
{
"epoch": 2.83,
"learning_rate": 2.0626664486032403e-05,
"loss": 2.1509,
"step": 1130
},
{
"epoch": 2.85,
"learning_rate": 1.9835937459564064e-05,
"loss": 2.1534,
"step": 1140
},
{
"epoch": 2.88,
"learning_rate": 1.905689761439075e-05,
"loss": 2.1388,
"step": 1150
},
{
"epoch": 2.9,
"learning_rate": 1.8289846792477393e-05,
"loss": 2.1408,
"step": 1160
},
{
"epoch": 2.93,
"learning_rate": 1.7535082190595988e-05,
"loss": 2.114,
"step": 1170
},
{
"epoch": 2.95,
"learning_rate": 1.6792896245175695e-05,
"loss": 2.1328,
"step": 1180
},
{
"epoch": 2.98,
"learning_rate": 1.6063576518997105e-05,
"loss": 2.1388,
"step": 1190
},
{
"epoch": 3.0,
"learning_rate": 1.5418423038143575e-05,
"loss": 2.1318,
"step": 1200
},
{
"epoch": 3.03,
"learning_rate": 1.4784122622824281e-05,
"loss": 2.1197,
"step": 1210
},
{
"epoch": 3.05,
"learning_rate": 1.4092315293469582e-05,
"loss": 2.0991,
"step": 1220
},
{
"epoch": 3.08,
"learning_rate": 1.341442053376587e-05,
"loss": 2.0859,
"step": 1230
},
{
"epoch": 3.1,
"learning_rate": 1.275070099662815e-05,
"loss": 2.0726,
"step": 1240
},
{
"epoch": 3.13,
"learning_rate": 1.2101413842727345e-05,
"loss": 2.1286,
"step": 1250
},
{
"epoch": 3.15,
"learning_rate": 1.146681064085231e-05,
"loss": 2.1482,
"step": 1260
},
{
"epoch": 3.18,
"learning_rate": 1.0847137270438563e-05,
"loss": 2.0831,
"step": 1270
},
{
"epoch": 3.2,
"learning_rate": 1.0242633826301368e-05,
"loss": 2.1293,
"step": 1280
},
{
"epoch": 3.23,
"learning_rate": 9.653534525610135e-06,
"loss": 2.1131,
"step": 1290
},
{
"epoch": 3.25,
"learning_rate": 9.080067617139982e-06,
"loss": 2.1132,
"step": 1300
},
{
"epoch": 3.28,
"learning_rate": 8.522455292835934e-06,
"loss": 2.1105,
"step": 1310
},
{
"epoch": 3.3,
"learning_rate": 7.98091360172381e-06,
"loss": 2.1572,
"step": 1320
},
{
"epoch": 3.33,
"learning_rate": 7.455652366201205e-06,
"loss": 2.1309,
"step": 1330
},
{
"epoch": 3.35,
"learning_rate": 6.946875100740946e-06,
"loss": 2.1169,
"step": 1340
},
{
"epoch": 3.38,
"learning_rate": 6.45477893303868e-06,
"loss": 2.1,
"step": 1350
},
{
"epoch": 3.4,
"learning_rate": 5.9795545276350115e-06,
"loss": 2.1159,
"step": 1360
},
{
"epoch": 3.43,
"learning_rate": 5.5664302371542185e-06,
"loss": 2.0945,
"step": 1370
},
{
"epoch": 3.45,
"learning_rate": 5.123763991028291e-06,
"loss": 2.1509,
"step": 1380
},
{
"epoch": 3.48,
"learning_rate": 4.6984852140263494e-06,
"loss": 2.1011,
"step": 1390
},
{
"epoch": 3.5,
"learning_rate": 4.290758682035634e-06,
"loss": 2.0756,
"step": 1400
},
{
"epoch": 3.53,
"learning_rate": 3.900742370258969e-06,
"loss": 2.0884,
"step": 1410
},
{
"epoch": 3.55,
"learning_rate": 3.5649949658293825e-06,
"loss": 2.1024,
"step": 1420
},
{
"epoch": 3.58,
"learning_rate": 3.20903867105079e-06,
"loss": 2.0449,
"step": 1430
},
{
"epoch": 3.6,
"learning_rate": 2.871211713038019e-06,
"loss": 2.1223,
"step": 1440
},
{
"epoch": 3.63,
"learning_rate": 2.551644984136853e-06,
"loss": 2.1164,
"step": 1450
},
{
"epoch": 3.65,
"learning_rate": 2.2504623016975535e-06,
"loss": 2.085,
"step": 1460
},
{
"epoch": 3.68,
"learning_rate": 1.9677803601013123e-06,
"loss": 2.0944,
"step": 1470
},
{
"epoch": 3.7,
"learning_rate": 1.70370868554659e-06,
"loss": 2.146,
"step": 1480
},
{
"epoch": 3.73,
"learning_rate": 1.4583495936127679e-06,
"loss": 2.0877,
"step": 1490
},
{
"epoch": 3.75,
"learning_rate": 1.2317981496176024e-06,
"loss": 2.1563,
"step": 1500
},
{
"epoch": 3.78,
"learning_rate": 1.0241421317837673e-06,
"loss": 2.1153,
"step": 1510
},
{
"epoch": 3.8,
"learning_rate": 8.354619972288702e-07,
"loss": 2.1226,
"step": 1520
},
{
"epoch": 3.83,
"learning_rate": 6.658308507920319e-07,
"loss": 2.0963,
"step": 1530
},
{
"epoch": 3.85,
"learning_rate": 5.153144167091395e-07,
"loss": 2.1122,
"step": 1540
},
{
"epoch": 3.88,
"learning_rate": 3.839710131477492e-07,
"loss": 2.1301,
"step": 1550
},
{
"epoch": 3.9,
"learning_rate": 2.927353201418681e-07,
"loss": 2.0882,
"step": 1560
},
{
"epoch": 3.93,
"learning_rate": 1.9602660834486252e-07,
"loss": 2.1299,
"step": 1570
},
{
"epoch": 3.95,
"learning_rate": 1.1861463635077785e-07,
"loss": 2.1095,
"step": 1580
},
{
"epoch": 3.98,
"learning_rate": 6.546751844955235e-08,
"loss": 2.0857,
"step": 1590
},
{
"epoch": 3.99,
"step": 1596,
"total_flos": 1.0918859302533857e+18,
"train_loss": 2.1782157116366507,
"train_runtime": 2104.4099,
"train_samples_per_second": 97.226,
"train_steps_per_second": 0.758
}
],
"max_steps": 1596,
"num_train_epochs": 4,
"total_flos": 1.0918859302533857e+18,
"trial_name": null,
"trial_params": null
}