777 lines
23 KiB
JSON
777 lines
23 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 1.995031055900621,
|
|
"eval_steps": 500,
|
|
"global_step": 536,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.037267080745341616,
|
|
"grad_norm": 0.19933472573757172,
|
|
"learning_rate": 2.8125000000000003e-05,
|
|
"loss": 3.1896,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 24.27871,
|
|
"step": 10,
|
|
"tokens/total": 368640,
|
|
"tokens/train_per_sec_per_gpu": 44.07,
|
|
"tokens/trainable": 321256
|
|
},
|
|
{
|
|
"epoch": 0.07453416149068323,
|
|
"grad_norm": 0.2485538274049759,
|
|
"learning_rate": 4.9995893860576395e-05,
|
|
"loss": 3.1828,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 24.11418,
|
|
"step": 20,
|
|
"tokens/total": 737280,
|
|
"tokens/train_per_sec_per_gpu": 43.47,
|
|
"tokens/trainable": 643308
|
|
},
|
|
{
|
|
"epoch": 0.11180124223602485,
|
|
"grad_norm": 0.16610555350780487,
|
|
"learning_rate": 4.99229333433282e-05,
|
|
"loss": 3.1795,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 24.03473,
|
|
"step": 30,
|
|
"tokens/total": 1105920,
|
|
"tokens/train_per_sec_per_gpu": 44.23,
|
|
"tokens/trainable": 962499
|
|
},
|
|
{
|
|
"epoch": 0.14906832298136646,
|
|
"grad_norm": 0.16600094735622406,
|
|
"learning_rate": 4.975903174511846e-05,
|
|
"loss": 3.1002,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.20239,
|
|
"step": 40,
|
|
"tokens/total": 1474560,
|
|
"tokens/train_per_sec_per_gpu": 45.06,
|
|
"tokens/trainable": 1283109
|
|
},
|
|
{
|
|
"epoch": 0.18633540372670807,
|
|
"grad_norm": 0.17242471873760223,
|
|
"learning_rate": 4.950478712510446e-05,
|
|
"loss": 3.0775,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.70407,
|
|
"step": 50,
|
|
"tokens/total": 1843200,
|
|
"tokens/train_per_sec_per_gpu": 45.42,
|
|
"tokens/trainable": 1603821
|
|
},
|
|
{
|
|
"epoch": 0.2236024844720497,
|
|
"grad_norm": 0.17550165951251984,
|
|
"learning_rate": 4.9161127194332655e-05,
|
|
"loss": 3.1334,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.95188,
|
|
"step": 60,
|
|
"tokens/total": 2211840,
|
|
"tokens/train_per_sec_per_gpu": 43.85,
|
|
"tokens/trainable": 1925471
|
|
},
|
|
{
|
|
"epoch": 0.2608695652173913,
|
|
"grad_norm": 0.18663033843040466,
|
|
"learning_rate": 4.872930593062168e-05,
|
|
"loss": 3.1937,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 24.37846,
|
|
"step": 70,
|
|
"tokens/total": 2580480,
|
|
"tokens/train_per_sec_per_gpu": 44.7,
|
|
"tokens/trainable": 2242912
|
|
},
|
|
{
|
|
"epoch": 0.2981366459627329,
|
|
"grad_norm": 0.20098601281642914,
|
|
"learning_rate": 4.8210899002933235e-05,
|
|
"loss": 3.1263,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.7895,
|
|
"step": 80,
|
|
"tokens/total": 2949120,
|
|
"tokens/train_per_sec_per_gpu": 44.38,
|
|
"tokens/trainable": 2562908
|
|
},
|
|
{
|
|
"epoch": 0.33540372670807456,
|
|
"grad_norm": 0.18968884646892548,
|
|
"learning_rate": 4.760779802192698e-05,
|
|
"loss": 3.142,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 23.15012,
|
|
"step": 90,
|
|
"tokens/total": 3317760,
|
|
"tokens/train_per_sec_per_gpu": 46.02,
|
|
"tokens/trainable": 2881497
|
|
},
|
|
{
|
|
"epoch": 0.37267080745341613,
|
|
"grad_norm": 0.21494834125041962,
|
|
"learning_rate": 4.692220363767833e-05,
|
|
"loss": 3.1794,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 24.03233,
|
|
"step": 100,
|
|
"tokens/total": 3686400,
|
|
"tokens/train_per_sec_per_gpu": 44.95,
|
|
"tokens/trainable": 3199097
|
|
},
|
|
{
|
|
"epoch": 0.40993788819875776,
|
|
"grad_norm": 0.20388352870941162,
|
|
"learning_rate": 4.6156617509745015e-05,
|
|
"loss": 3.1358,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 23.00703,
|
|
"step": 110,
|
|
"tokens/total": 4055040,
|
|
"tokens/train_per_sec_per_gpu": 44.96,
|
|
"tokens/trainable": 3517985
|
|
},
|
|
{
|
|
"epoch": 0.4472049689440994,
|
|
"grad_norm": 0.22123299539089203,
|
|
"learning_rate": 4.531383317888261e-05,
|
|
"loss": 3.1032,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.2691,
|
|
"step": 120,
|
|
"tokens/total": 4423680,
|
|
"tokens/train_per_sec_per_gpu": 41.16,
|
|
"tokens/trainable": 3837324
|
|
},
|
|
{
|
|
"epoch": 0.484472049689441,
|
|
"grad_norm": 0.2072950154542923,
|
|
"learning_rate": 4.439692587371731e-05,
|
|
"loss": 3.1074,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.36283,
|
|
"step": 130,
|
|
"tokens/total": 4792320,
|
|
"tokens/train_per_sec_per_gpu": 46.1,
|
|
"tokens/trainable": 4159485
|
|
},
|
|
{
|
|
"epoch": 0.5217391304347826,
|
|
"grad_norm": 0.22470185160636902,
|
|
"learning_rate": 4.3409241289570156e-05,
|
|
"loss": 3.0826,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.81505,
|
|
"step": 140,
|
|
"tokens/total": 5160960,
|
|
"tokens/train_per_sec_per_gpu": 45.04,
|
|
"tokens/trainable": 4479023
|
|
},
|
|
{
|
|
"epoch": 0.5590062111801242,
|
|
"grad_norm": 0.23675978183746338,
|
|
"learning_rate": 4.235438338037764e-05,
|
|
"loss": 3.0895,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.96609,
|
|
"step": 150,
|
|
"tokens/total": 5529600,
|
|
"tokens/train_per_sec_per_gpu": 44.54,
|
|
"tokens/trainable": 4798599
|
|
},
|
|
{
|
|
"epoch": 0.5962732919254659,
|
|
"grad_norm": 0.22080311179161072,
|
|
"learning_rate": 4.123620120825459e-05,
|
|
"loss": 3.0991,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.17798,
|
|
"step": 160,
|
|
"tokens/total": 5898240,
|
|
"tokens/train_per_sec_per_gpu": 45.18,
|
|
"tokens/trainable": 5117093
|
|
},
|
|
{
|
|
"epoch": 0.6335403726708074,
|
|
"grad_norm": 0.22512708604335785,
|
|
"learning_rate": 4.005877489868401e-05,
|
|
"loss": 3.1219,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.68945,
|
|
"step": 170,
|
|
"tokens/total": 6266880,
|
|
"tokens/train_per_sec_per_gpu": 43.85,
|
|
"tokens/trainable": 5436542
|
|
},
|
|
{
|
|
"epoch": 0.6708074534161491,
|
|
"grad_norm": 0.22725142538547516,
|
|
"learning_rate": 3.88264007525816e-05,
|
|
"loss": 3.167,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 23.73617,
|
|
"step": 180,
|
|
"tokens/total": 6635520,
|
|
"tokens/train_per_sec_per_gpu": 45.1,
|
|
"tokens/trainable": 5757152
|
|
},
|
|
{
|
|
"epoch": 0.7080745341614907,
|
|
"grad_norm": 0.23717202246189117,
|
|
"learning_rate": 3.754357556955966e-05,
|
|
"loss": 3.1361,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 23.01394,
|
|
"step": 190,
|
|
"tokens/total": 7004160,
|
|
"tokens/train_per_sec_per_gpu": 42.08,
|
|
"tokens/trainable": 6076610
|
|
},
|
|
{
|
|
"epoch": 0.7453416149068323,
|
|
"grad_norm": 0.2373926043510437,
|
|
"learning_rate": 3.621498023959294e-05,
|
|
"loss": 3.1262,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.78722,
|
|
"step": 200,
|
|
"tokens/total": 7372800,
|
|
"tokens/train_per_sec_per_gpu": 44.68,
|
|
"tokens/trainable": 6396405
|
|
},
|
|
{
|
|
"epoch": 0.782608695652174,
|
|
"grad_norm": 0.22468852996826172,
|
|
"learning_rate": 3.484546266295883e-05,
|
|
"loss": 3.1138,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.50641,
|
|
"step": 210,
|
|
"tokens/total": 7741440,
|
|
"tokens/train_per_sec_per_gpu": 43.14,
|
|
"tokens/trainable": 6714000
|
|
},
|
|
{
|
|
"epoch": 0.8198757763975155,
|
|
"grad_norm": 0.23512248694896698,
|
|
"learning_rate": 3.344002006077487e-05,
|
|
"loss": 3.113,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.48841,
|
|
"step": 220,
|
|
"tokens/total": 8110080,
|
|
"tokens/train_per_sec_per_gpu": 44.86,
|
|
"tokens/trainable": 7034597
|
|
},
|
|
{
|
|
"epoch": 0.8571428571428571,
|
|
"grad_norm": 0.23428736627101898,
|
|
"learning_rate": 3.200378074068092e-05,
|
|
"loss": 3.0454,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.01844,
|
|
"step": 230,
|
|
"tokens/total": 8478720,
|
|
"tokens/train_per_sec_per_gpu": 44.72,
|
|
"tokens/trainable": 7357866
|
|
},
|
|
{
|
|
"epoch": 0.8944099378881988,
|
|
"grad_norm": 0.25676628947257996,
|
|
"learning_rate": 3.0541985384200686e-05,
|
|
"loss": 3.1156,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.54695,
|
|
"step": 240,
|
|
"tokens/total": 8847360,
|
|
"tokens/train_per_sec_per_gpu": 44.71,
|
|
"tokens/trainable": 7675991
|
|
},
|
|
{
|
|
"epoch": 0.9316770186335404,
|
|
"grad_norm": 0.2449398934841156,
|
|
"learning_rate": 2.9059967924063153e-05,
|
|
"loss": 3.1204,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.65544,
|
|
"step": 250,
|
|
"tokens/total": 9216000,
|
|
"tokens/train_per_sec_per_gpu": 45.14,
|
|
"tokens/trainable": 7995656
|
|
},
|
|
{
|
|
"epoch": 0.968944099378882,
|
|
"grad_norm": 0.2541828453540802,
|
|
"learning_rate": 2.7563136081260582e-05,
|
|
"loss": 3.1261,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.78494,
|
|
"step": 260,
|
|
"tokens/total": 9584640,
|
|
"tokens/train_per_sec_per_gpu": 42.8,
|
|
"tokens/trainable": 8314538
|
|
},
|
|
{
|
|
"epoch": 1.0037267080745342,
|
|
"grad_norm": 0.24245300889015198,
|
|
"learning_rate": 2.6056951632860976e-05,
|
|
"loss": 3.0867,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.90467,
|
|
"step": 270,
|
|
"tokens/total": 9924608,
|
|
"tokens/train_per_sec_per_gpu": 45.48,
|
|
"tokens/trainable": 8609040
|
|
},
|
|
{
|
|
"epoch": 1.0409937888198757,
|
|
"grad_norm": 0.2593900263309479,
|
|
"learning_rate": 2.4546910482575908e-05,
|
|
"loss": 3.0741,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.63041,
|
|
"step": 280,
|
|
"tokens/total": 10293248,
|
|
"tokens/train_per_sec_per_gpu": 44.2,
|
|
"tokens/trainable": 8927894
|
|
},
|
|
{
|
|
"epoch": 1.0782608695652174,
|
|
"grad_norm": 0.2553490996360779,
|
|
"learning_rate": 2.303852260680388e-05,
|
|
"loss": 3.0859,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.88716,
|
|
"step": 290,
|
|
"tokens/total": 10661888,
|
|
"tokens/train_per_sec_per_gpu": 46.62,
|
|
"tokens/trainable": 9250881
|
|
},
|
|
{
|
|
"epoch": 1.115527950310559,
|
|
"grad_norm": 0.2889447510242462,
|
|
"learning_rate": 2.1537291949324007e-05,
|
|
"loss": 3.0993,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.18242,
|
|
"step": 300,
|
|
"tokens/total": 11030528,
|
|
"tokens/train_per_sec_per_gpu": 43.46,
|
|
"tokens/trainable": 9569197
|
|
},
|
|
{
|
|
"epoch": 1.1527950310559005,
|
|
"grad_norm": 0.3012157380580902,
|
|
"learning_rate": 2.0048696338002015e-05,
|
|
"loss": 3.1172,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.58306,
|
|
"step": 310,
|
|
"tokens/total": 11399168,
|
|
"tokens/train_per_sec_per_gpu": 42.07,
|
|
"tokens/trainable": 9889854
|
|
},
|
|
{
|
|
"epoch": 1.1900621118012422,
|
|
"grad_norm": 0.290158212184906,
|
|
"learning_rate": 1.857816749679043e-05,
|
|
"loss": 3.1,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.19795,
|
|
"step": 320,
|
|
"tokens/total": 11767808,
|
|
"tokens/train_per_sec_per_gpu": 43.72,
|
|
"tokens/trainable": 10207408
|
|
},
|
|
{
|
|
"epoch": 1.227329192546584,
|
|
"grad_norm": 0.315685898065567,
|
|
"learning_rate": 1.713107122595693e-05,
|
|
"loss": 3.0941,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.06737,
|
|
"step": 330,
|
|
"tokens/total": 12136448,
|
|
"tokens/train_per_sec_per_gpu": 43.89,
|
|
"tokens/trainable": 10527550
|
|
},
|
|
{
|
|
"epoch": 1.2645962732919256,
|
|
"grad_norm": 0.29844436049461365,
|
|
"learning_rate": 1.571268782286107e-05,
|
|
"loss": 3.0997,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.19129,
|
|
"step": 340,
|
|
"tokens/total": 12505088,
|
|
"tokens/train_per_sec_per_gpu": 45.94,
|
|
"tokens/trainable": 10848911
|
|
},
|
|
{
|
|
"epoch": 1.301863354037267,
|
|
"grad_norm": 0.29774603247642517,
|
|
"learning_rate": 1.4328192814721846e-05,
|
|
"loss": 3.0617,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.36384,
|
|
"step": 350,
|
|
"tokens/total": 12873728,
|
|
"tokens/train_per_sec_per_gpu": 44.22,
|
|
"tokens/trainable": 11167329
|
|
},
|
|
{
|
|
"epoch": 1.3391304347826087,
|
|
"grad_norm": 0.305044025182724,
|
|
"learning_rate": 1.2982638073680112e-05,
|
|
"loss": 3.0857,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.88278,
|
|
"step": 360,
|
|
"tokens/total": 13242368,
|
|
"tokens/train_per_sec_per_gpu": 44.71,
|
|
"tokens/trainable": 11489331
|
|
},
|
|
{
|
|
"epoch": 1.3763975155279504,
|
|
"grad_norm": 0.30985885858535767,
|
|
"learning_rate": 1.1680933383065057e-05,
|
|
"loss": 3.0768,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.68889,
|
|
"step": 370,
|
|
"tokens/total": 13611008,
|
|
"tokens/train_per_sec_per_gpu": 43.03,
|
|
"tokens/trainable": 11809342
|
|
},
|
|
{
|
|
"epoch": 1.4136645962732919,
|
|
"grad_norm": 0.31623899936676025,
|
|
"learning_rate": 1.042782852212725e-05,
|
|
"loss": 3.0995,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.18686,
|
|
"step": 380,
|
|
"tokens/total": 13979648,
|
|
"tokens/train_per_sec_per_gpu": 45.18,
|
|
"tokens/trainable": 12127653
|
|
},
|
|
{
|
|
"epoch": 1.4509316770186336,
|
|
"grad_norm": 0.31580615043640137,
|
|
"learning_rate": 9.227895934609235e-06,
|
|
"loss": 3.0674,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.48597,
|
|
"step": 390,
|
|
"tokens/total": 14348288,
|
|
"tokens/train_per_sec_per_gpu": 45.49,
|
|
"tokens/trainable": 12446447
|
|
},
|
|
{
|
|
"epoch": 1.4881987577639753,
|
|
"grad_norm": 0.3268173635005951,
|
|
"learning_rate": 8.085514044394216e-06,
|
|
"loss": 3.1286,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.84198,
|
|
"step": 400,
|
|
"tokens/total": 14716928,
|
|
"tokens/train_per_sec_per_gpu": 44.11,
|
|
"tokens/trainable": 12766899
|
|
},
|
|
{
|
|
"epoch": 1.5254658385093167,
|
|
"grad_norm": 0.31086301803588867,
|
|
"learning_rate": 7.004851279112129e-06,
|
|
"loss": 3.0784,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.72362,
|
|
"step": 410,
|
|
"tokens/total": 15085568,
|
|
"tokens/train_per_sec_per_gpu": 45.54,
|
|
"tokens/trainable": 13088335
|
|
},
|
|
{
|
|
"epoch": 1.5627329192546584,
|
|
"grad_norm": 0.3192853629589081,
|
|
"learning_rate": 5.989850859999227e-06,
|
|
"loss": 3.0708,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.55914,
|
|
"step": 420,
|
|
"tokens/total": 15454208,
|
|
"tokens/train_per_sec_per_gpu": 44.24,
|
|
"tokens/trainable": 13408321
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"grad_norm": 0.3102209270000458,
|
|
"learning_rate": 5.044216413511494e-06,
|
|
"loss": 3.0687,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.51392,
|
|
"step": 430,
|
|
"tokens/total": 15822848,
|
|
"tokens/train_per_sec_per_gpu": 44.8,
|
|
"tokens/trainable": 13731661
|
|
},
|
|
{
|
|
"epoch": 1.6372670807453416,
|
|
"grad_norm": 0.3230491876602173,
|
|
"learning_rate": 4.171398457193274e-06,
|
|
"loss": 3.0735,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.61743,
|
|
"step": 440,
|
|
"tokens/total": 16191488,
|
|
"tokens/train_per_sec_per_gpu": 45.26,
|
|
"tokens/trainable": 14052955
|
|
},
|
|
{
|
|
"epoch": 1.6745341614906832,
|
|
"grad_norm": 0.3240269124507904,
|
|
"learning_rate": 3.3745818091130033e-06,
|
|
"loss": 3.1082,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.38072,
|
|
"step": 450,
|
|
"tokens/total": 16560128,
|
|
"tokens/train_per_sec_per_gpu": 43.83,
|
|
"tokens/trainable": 14372424
|
|
},
|
|
{
|
|
"epoch": 1.711801242236025,
|
|
"grad_norm": 0.3217725157737732,
|
|
"learning_rate": 2.6566739668074576e-06,
|
|
"loss": 3.1315,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.90832,
|
|
"step": 460,
|
|
"tokens/total": 16928768,
|
|
"tokens/train_per_sec_per_gpu": 44.5,
|
|
"tokens/trainable": 14690565
|
|
},
|
|
{
|
|
"epoch": 1.7490683229813664,
|
|
"grad_norm": 0.3098929524421692,
|
|
"learning_rate": 2.020294498138503e-06,
|
|
"loss": 3.066,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.45591,
|
|
"step": 470,
|
|
"tokens/total": 17297408,
|
|
"tokens/train_per_sec_per_gpu": 45.19,
|
|
"tokens/trainable": 15008561
|
|
},
|
|
{
|
|
"epoch": 1.786335403726708,
|
|
"grad_norm": 0.3308022618293762,
|
|
"learning_rate": 1.4677654827741111e-06,
|
|
"loss": 3.1263,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.7895,
|
|
"step": 480,
|
|
"tokens/total": 17666048,
|
|
"tokens/train_per_sec_per_gpu": 44.69,
|
|
"tokens/trainable": 15325346
|
|
},
|
|
{
|
|
"epoch": 1.8236024844720498,
|
|
"grad_norm": 0.33477458357810974,
|
|
"learning_rate": 1.0011030391714278e-06,
|
|
"loss": 3.0856,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.88059,
|
|
"step": 490,
|
|
"tokens/total": 18034688,
|
|
"tokens/train_per_sec_per_gpu": 44.1,
|
|
"tokens/trainable": 15644995
|
|
},
|
|
{
|
|
"epoch": 1.8608695652173912,
|
|
"grad_norm": 0.3346908390522003,
|
|
"learning_rate": 6.220099679791897e-07,
|
|
"loss": 3.1043,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.29361,
|
|
"step": 500,
|
|
"tokens/total": 18403328,
|
|
"tokens/train_per_sec_per_gpu": 44.87,
|
|
"tokens/trainable": 15966250
|
|
},
|
|
{
|
|
"epoch": 1.898136645962733,
|
|
"grad_norm": 0.3268662095069885,
|
|
"learning_rate": 3.3186953870281144e-07,
|
|
"loss": 3.0731,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 21.60879,
|
|
"step": 510,
|
|
"tokens/total": 18771968,
|
|
"tokens/train_per_sec_per_gpu": 44.17,
|
|
"tokens/trainable": 16285364
|
|
},
|
|
{
|
|
"epoch": 1.9354037267080746,
|
|
"grad_norm": 0.32842233777046204,
|
|
"learning_rate": 1.3174044230396532e-07,
|
|
"loss": 3.1173,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.58532,
|
|
"step": 520,
|
|
"tokens/total": 19140608,
|
|
"tokens/train_per_sec_per_gpu": 42.71,
|
|
"tokens/trainable": 16603401
|
|
},
|
|
{
|
|
"epoch": 1.972670807453416,
|
|
"grad_norm": 0.33898642659187317,
|
|
"learning_rate": 2.2352928152089003e-08,
|
|
"loss": 3.1123,
|
|
"memory/device_reserved (GiB)": 45.16,
|
|
"memory/max_active (GiB)": 37.89,
|
|
"memory/max_allocated (GiB)": 37.89,
|
|
"ppl": 22.47267,
|
|
"step": 530,
|
|
"tokens/total": 19509248,
|
|
"tokens/train_per_sec_per_gpu": 44.84,
|
|
"tokens/trainable": 16923422
|
|
}
|
|
],
|
|
"logging_steps": 10,
|
|
"max_steps": 536,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 2,
|
|
"save_steps": 302,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 8.394277834016686e+17,
|
|
"train_batch_size": 6,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|