Files
smollm2-135m-noval/last-checkpoint/trainer_state.json

3622 lines
94 KiB
JSON
Raw Normal View History

{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.303500423701193,
"eval_steps": 500,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0002607391956195815,
"grad_norm": 1.049068570137024,
"learning_rate": 0.0,
"loss": 2.1770923137664795,
"step": 1
},
{
"epoch": 0.002607391956195815,
"grad_norm": 0.7216854691505432,
"learning_rate": 2.2499999999999998e-05,
"loss": 2.2134272257486978,
"step": 10
},
{
"epoch": 0.00521478391239163,
"grad_norm": 0.6660354733467102,
"learning_rate": 4.75e-05,
"loss": 2.135795021057129,
"step": 20
},
{
"epoch": 0.007822175868587446,
"grad_norm": 0.474956214427948,
"learning_rate": 7.25e-05,
"loss": 1.944338607788086,
"step": 30
},
{
"epoch": 0.01042956782478326,
"grad_norm": 0.4872700870037079,
"learning_rate": 9.750000000000001e-05,
"loss": 1.7077013015747071,
"step": 40
},
{
"epoch": 0.013036959780979076,
"grad_norm": 0.45002982020378113,
"learning_rate": 0.0001225,
"loss": 1.567308521270752,
"step": 50
},
{
"epoch": 0.01564435173717489,
"grad_norm": 0.4571980834007263,
"learning_rate": 0.0001475,
"loss": 1.602952003479004,
"step": 60
},
{
"epoch": 0.018251743693370705,
"grad_norm": 0.4689195156097412,
"learning_rate": 0.0001725,
"loss": 1.5603692054748535,
"step": 70
},
{
"epoch": 0.02085913564956652,
"grad_norm": 0.33675292134284973,
"learning_rate": 0.0001975,
"loss": 1.5227128982543945,
"step": 80
},
{
"epoch": 0.023466527605762337,
"grad_norm": 0.44045963883399963,
"learning_rate": 0.00022250000000000001,
"loss": 1.4949231147766113,
"step": 90
},
{
"epoch": 0.02607391956195815,
"grad_norm": 0.5103346705436707,
"learning_rate": 0.0002475,
"loss": 1.4100960731506347,
"step": 100
},
{
"epoch": 0.028681311518153965,
"grad_norm": 0.3978748321533203,
"learning_rate": 0.0002725,
"loss": 1.4386549949645997,
"step": 110
},
{
"epoch": 0.03128870347434978,
"grad_norm": 0.46545708179473877,
"learning_rate": 0.00029749999999999997,
"loss": 1.4639436721801757,
"step": 120
},
{
"epoch": 0.03389609543054559,
"grad_norm": 0.39714980125427246,
"learning_rate": 0.00032250000000000003,
"loss": 1.4476322174072265,
"step": 130
},
{
"epoch": 0.03650348738674141,
"grad_norm": 0.4530780017375946,
"learning_rate": 0.0003475,
"loss": 1.4174782752990722,
"step": 140
},
{
"epoch": 0.03911087934293723,
"grad_norm": 0.4009428322315216,
"learning_rate": 0.0003725,
"loss": 1.3690480232238769,
"step": 150
},
{
"epoch": 0.04171827129913304,
"grad_norm": 0.41648587584495544,
"learning_rate": 0.0003975,
"loss": 1.3331147193908692,
"step": 160
},
{
"epoch": 0.04432566325532886,
"grad_norm": 0.46143171191215515,
"learning_rate": 0.00042249999999999997,
"loss": 1.3242422103881837,
"step": 170
},
{
"epoch": 0.046933055211524674,
"grad_norm": 0.5419852137565613,
"learning_rate": 0.00044750000000000004,
"loss": 1.3132104873657227,
"step": 180
},
{
"epoch": 0.049540447167720485,
"grad_norm": 0.4304125905036926,
"learning_rate": 0.0004725,
"loss": 1.3139053344726563,
"step": 190
},
{
"epoch": 0.0521478391239163,
"grad_norm": 0.39391469955444336,
"learning_rate": 0.0004975,
"loss": 1.3622180938720703,
"step": 200
},
{
"epoch": 0.05475523108011212,
"grad_norm": 0.4835835099220276,
"learning_rate": 0.0004999956627840445,
"loss": 1.2926831245422363,
"step": 210
},
{
"epoch": 0.05736262303630793,
"grad_norm": 0.3858627378940582,
"learning_rate": 0.0004999806701319743,
"loss": 1.2902835845947265,
"step": 220
},
{
"epoch": 0.05997001499250375,
"grad_norm": 0.40019169449806213,
"learning_rate": 0.000499954969139996,
"loss": 1.255649948120117,
"step": 230
},
{
"epoch": 0.06257740694869957,
"grad_norm": 0.3915915787220001,
"learning_rate": 0.0004999185609090544,
"loss": 1.2121038436889648,
"step": 240
},
{
"epoch": 0.06518479890489538,
"grad_norm": 0.4412119686603546,
"learning_rate": 0.0004998714469987571,
"loss": 1.3129422187805175,
"step": 250
},
{
"epoch": 0.06779219086109119,
"grad_norm": 0.37609967589378357,
"learning_rate": 0.000499813629427307,
"loss": 1.165921974182129,
"step": 260
},
{
"epoch": 0.070399582817287,
"grad_norm": 0.3630351722240448,
"learning_rate": 0.0004997451106714166,
"loss": 1.233005905151367,
"step": 270
},
{
"epoch": 0.07300697477348282,
"grad_norm": 0.3163871467113495,
"learning_rate": 0.0004996658936662013,
"loss": 1.2466936111450195,
"step": 280
},
{
"epoch": 0.07561436672967864,
"grad_norm": 0.3848457932472229,
"learning_rate": 0.000499575981805054,
"loss": 1.2024270057678224,
"step": 290
},
{
"epoch": 0.07822175868587446,
"grad_norm": 0.4009913206100464,
"learning_rate": 0.0004994753789394994,
"loss": 1.2530613899230958,
"step": 300
},
{
"epoch": 0.08082915064207027,
"grad_norm": 0.3757595121860504,
"learning_rate": 0.0004993640893790298,
"loss": 1.2115967750549317,
"step": 310
},
{
"epoch": 0.08343654259826608,
"grad_norm": 0.43078887462615967,
"learning_rate": 0.0004992421178909191,
"loss": 1.1773183822631836,
"step": 320
},
{
"epoch": 0.0860439345544619,
"grad_norm": 0.33781853318214417,
"learning_rate": 0.0004991094697000202,
"loss": 1.257601261138916,
"step": 330
},
{
"epoch": 0.08865132651065771,
"grad_norm": 0.39425569772720337,
"learning_rate": 0.00049896615048854,
"loss": 1.208796501159668,
"step": 340
},
{
"epoch": 0.09125871846685353,
"grad_norm": 0.7240598201751709,
"learning_rate": 0.0004988121663957966,
"loss": 1.1153605461120606,
"step": 350
},
{
"epoch": 0.09386611042304935,
"grad_norm": 0.3594074845314026,
"learning_rate": 0.0004986475240179559,
"loss": 1.1370932579040527,
"step": 360
},
{
"epoch": 0.09647350237924517,
"grad_norm": 0.3602915406227112,
"learning_rate": 0.0004984722304077496,
"loss": 1.2043560981750487,
"step": 370
},
{
"epoch": 0.09908089433544097,
"grad_norm": 0.4028254747390747,
"learning_rate": 0.0004982862930741725,
"loss": 1.1967852592468262,
"step": 380
},
{
"epoch": 0.10168828629163679,
"grad_norm": 0.3369152247905731,
"learning_rate": 0.0004980897199821609,
"loss": 1.2563385009765624,
"step": 390
},
{
"epoch": 0.1042956782478326,
"grad_norm": 0.3429180979728699,
"learning_rate": 0.000497882519552252,
"loss": 1.1891561508178712,
"step": 400
},
{
"epoch": 0.10690307020402842,
"grad_norm": 0.3755589425563812,
"learning_rate": 0.0004976647006602225,
"loss": 1.214197826385498,
"step": 410
},
{
"epoch": 0.10951046216022424,
"grad_norm": 0.34570881724357605,
"learning_rate": 0.0004974362726367089,
"loss": 1.1685050010681153,
"step": 420
},
{
"epoch": 0.11211785411642006,
"grad_norm": 0.33463919162750244,
"learning_rate": 0.0004971972452668074,
"loss": 1.1726037979125976,
"step": 430
},
{
"epoch": 0.11472524607261586,
"grad_norm": 0.3386293053627014,
"learning_rate": 0.000496947628789655,
"loss": 1.0812938690185547,
"step": 440
},
{
"epoch": 0.11733263802881168,
"grad_norm": 0.37149256467819214,
"learning_rate": 0.0004966874338979907,
"loss": 1.1715264320373535,
"step": 450
},
{
"epoch": 0.1199400299850075,
"grad_norm": 0.33630073070526123,
"learning_rate": 0.0004964166717376978,
"loss": 1.165538215637207,
"step": 460
},
{
"epoch": 0.12254742194120331,
"grad_norm": 0.37086284160614014,
"learning_rate": 0.0004961353539073258,
"loss": 1.0824301719665528,
"step": 470
},
{
"epoch": 0.12515481389739913,
"grad_norm": 0.2987000048160553,
"learning_rate": 0.0004958434924575947,
"loss": 1.1530218124389648,
"step": 480
},
{
"epoch": 0.12776220585359493,
"grad_norm": 0.30463266372680664,
"learning_rate": 0.0004955410998908774,
"loss": 1.1471798896789551,
"step": 490
},
{
"epoch": 0.13036959780979077,
"grad_norm": 0.32853904366493225,
"learning_rate": 0.0004952281891606649,
"loss": 1.0868051528930665,
"step": 500
},
{
"epoch": 0.13036959780979077,
"eval_loss": 2.1237475872039795,
"eval_runtime": 60.7937,
"eval_samples_per_second": 9.738,
"eval_steps_per_second": 1.217,
"step": 500
},
{
"epoch": 0.13297698976598657,
"grad_norm": 0.4235430955886841,
"learning_rate": 0.0004949047736710116,
"loss": 1.1696773529052735,
"step": 510
},
{
"epoch": 0.13558438172218237,
"grad_norm": 0.36118283867836,
"learning_rate": 0.0004945708672759606,
"loss": 1.0913918495178223,
"step": 520
},
{
"epoch": 0.1381917736783782,
"grad_norm": 0.33532798290252686,
"learning_rate": 0.0004942264842789506,
"loss": 1.1192610740661622,
"step": 530
},
{
"epoch": 0.140799165634574,
"grad_norm": 0.34322044253349304,
"learning_rate": 0.0004938716394322028,
"loss": 1.248635482788086,
"step": 540
},
{
"epoch": 0.14340655759076984,
"grad_norm": 0.3750886917114258,
"learning_rate": 0.0004935063479360897,
"loss": 1.169818115234375,
"step": 550
},
{
"epoch": 0.14601394954696564,
"grad_norm": 0.3408745229244232,
"learning_rate": 0.000493130625438483,
"loss": 1.090888214111328,
"step": 560
},
{
"epoch": 0.14862134150316147,
"grad_norm": 0.3277114927768707,
"learning_rate": 0.000492744488034084,
"loss": 1.1033119201660155,
"step": 570
},
{
"epoch": 0.15122873345935728,
"grad_norm": 0.2922538220882416,
"learning_rate": 0.0004923479522637341,
"loss": 1.0497876167297364,
"step": 580
},
{
"epoch": 0.15383612541555308,
"grad_norm": 0.4220677614212036,
"learning_rate": 0.0004919410351137058,
"loss": 1.0967654228210448,
"step": 590
},
{
"epoch": 0.1564435173717489,
"grad_norm": 0.35899877548217773,
"learning_rate": 0.0004915237540149755,
"loss": 1.0839926719665527,
"step": 600
},
{
"epoch": 0.15905090932794472,
"grad_norm": 0.3528715670108795,
"learning_rate": 0.0004910961268424766,
"loss": 1.071396541595459,
"step": 610
},
{
"epoch": 0.16165830128414055,
"grad_norm": 0.3362886905670166,
"learning_rate": 0.000490658171914334,
"loss": 1.1138338088989257,
"step": 620
},
{
"epoch": 0.16426569324033635,
"grad_norm": 0.35801777243614197,
"learning_rate": 0.0004902099079910794,
"loss": 1.0978638648986816,
"step": 630
},
{
"epoch": 0.16687308519653216,
"grad_norm": 0.4060690402984619,
"learning_rate": 0.0004897513542748468,
"loss": 1.0912402153015137,
"step": 640
},
{
"epoch": 0.169480477152728,
"grad_norm": 0.3944070041179657,
"learning_rate": 0.0004892825304085517,
"loss": 1.1369976997375488,
"step": 650
},
{
"epoch": 0.1720878691089238,
"grad_norm": 0.34240999817848206,
"learning_rate": 0.000488803456475048,
"loss": 1.0700874328613281,
"step": 660
},
{
"epoch": 0.17469526106511962,
"grad_norm": 0.3258955776691437,
"learning_rate": 0.0004883141529962683,
"loss": 1.1293649673461914,
"step": 670
},
{
"epoch": 0.17730265302131543,
"grad_norm": 0.32214921712875366,
"learning_rate": 0.0004878146409323456,
"loss": 0.9946167945861817,
"step": 680
},
{
"epoch": 0.17991004497751126,
"grad_norm": 0.3059631884098053,
"learning_rate": 0.0004873049416807138,
"loss": 1.0654255867004394,
"step": 690
},
{
"epoch": 0.18251743693370706,
"grad_norm": 0.27094215154647827,
"learning_rate": 0.0004867850770751926,
"loss": 1.03268461227417,
"step": 700
},
{
"epoch": 0.18512482888990286,
"grad_norm": 0.3041401207447052,
"learning_rate": 0.00048625506938505136,
"loss": 1.0983957290649413,
"step": 710
},
{
"epoch": 0.1877322208460987,
"grad_norm": 0.2661817967891693,
"learning_rate": 0.00048571494131405567,
"loss": 1.051801586151123,
"step": 720
},
{
"epoch": 0.1903396128022945,
"grad_norm": 0.36525386571884155,
"learning_rate": 0.0004851647159994943,
"loss": 1.1403950691223144,
"step": 730
},
{
"epoch": 0.19294700475849033,
"grad_norm": 0.37239551544189453,
"learning_rate": 0.0004846044170111884,
"loss": 1.0938661575317383,
"step": 740
},
{
"epoch": 0.19555439671468614,
"grad_norm": 0.29340049624443054,
"learning_rate": 0.0004840340683504819,
"loss": 1.030819034576416,
"step": 750
},
{
"epoch": 0.19816178867088194,
"grad_norm": 0.37587422132492065,
"learning_rate": 0.00048345369444921315,
"loss": 1.1062429428100586,
"step": 760
},
{
"epoch": 0.20076918062707777,
"grad_norm": 0.3974108397960663,
"learning_rate": 0.0004828633201686684,
"loss": 1.206178855895996,
"step": 770
},
{
"epoch": 0.20337657258327357,
"grad_norm": 0.2957328259944916,
"learning_rate": 0.0004822629707985169,
"loss": 1.0188568115234375,
"step": 780
},
{
"epoch": 0.2059839645394694,
"grad_norm": 0.333222895860672,
"learning_rate": 0.0004816526720557276,
"loss": 1.0218161582946776,
"step": 790
},
{
"epoch": 0.2085913564956652,
"grad_norm": 0.3079843521118164,
"learning_rate": 0.00048103245008346735,
"loss": 1.0325109481811523,
"step": 800
},
{
"epoch": 0.211198748451861,
"grad_norm": 0.3701516091823578,
"learning_rate": 0.00048040233144998123,
"loss": 1.0947632789611816,
"step": 810
},
{
"epoch": 0.21380614040805684,
"grad_norm": 0.3121979236602783,
"learning_rate": 0.0004797623431474543,
"loss": 1.0280736923217773,
"step": 820
},
{
"epoch": 0.21641353236425265,
"grad_norm": 0.2963069677352905,
"learning_rate": 0.0004791125125908554,
"loss": 1.0478859901428224,
"step": 830
},
{
"epoch": 0.21902092432044848,
"grad_norm": 0.2818042039871216,
"learning_rate": 0.00047845286761676276,
"loss": 1.0170639991760253,
"step": 840
},
{
"epoch": 0.22162831627664428,
"grad_norm": 0.6271336674690247,
"learning_rate": 0.00047778343648217155,
"loss": 1.0549551963806152,
"step": 850
},
{
"epoch": 0.22423570823284011,
"grad_norm": 0.39998874068260193,
"learning_rate": 0.0004771042478632836,
"loss": 0.9978870391845703,
"step": 860
},
{
"epoch": 0.22684310018903592,
"grad_norm": 0.3056093752384186,
"learning_rate": 0.0004764153308542788,
"loss": 1.0474911689758302,
"step": 870
},
{
"epoch": 0.22945049214523172,
"grad_norm": 0.36767828464508057,
"learning_rate": 0.00047571671496606893,
"loss": 0.9711828231811523,
"step": 880
},
{
"epoch": 0.23205788410142755,
"grad_norm": 0.3417050540447235,
"learning_rate": 0.0004750084301250335,
"loss": 0.9817624092102051,
"step": 890
},
{
"epoch": 0.23466527605762336,
"grad_norm": 0.4010846018791199,
"learning_rate": 0.0004742905066717377,
"loss": 1.0280840873718262,
"step": 900
},
{
"epoch": 0.2372726680138192,
"grad_norm": 0.3609777092933655,
"learning_rate": 0.0004735629753596328,
"loss": 0.9949556350708008,
"step": 910
},
{
"epoch": 0.239880059970015,
"grad_norm": 0.4105432629585266,
"learning_rate": 0.00047282586735373887,
"loss": 1.0682054519653321,
"step": 920
},
{
"epoch": 0.2424874519262108,
"grad_norm": 0.3331395387649536,
"learning_rate": 0.00047207921422930967,
"loss": 1.013453483581543,
"step": 930
},
{
"epoch": 0.24509484388240663,
"grad_norm": 0.4003639817237854,
"learning_rate": 0.00047132304797047975,
"loss": 0.9975796699523926,
"step": 940
},
{
"epoch": 0.24770223583860243,
"grad_norm": 0.28267544507980347,
"learning_rate": 0.00047055740096889516,
"loss": 0.9484305381774902,
"step": 950
},
{
"epoch": 0.25030962779479826,
"grad_norm": 0.32790929079055786,
"learning_rate": 0.00046978230602232507,
"loss": 1.0573930740356445,
"step": 960
},
{
"epoch": 0.25291701975099407,
"grad_norm": 0.30518388748168945,
"learning_rate": 0.0004689977963332572,
"loss": 0.9687243461608886,
"step": 970
},
{
"epoch": 0.25552441170718987,
"grad_norm": 0.31517553329467773,
"learning_rate": 0.00046820390550747585,
"loss": 0.9568646430969239,
"step": 980
},
{
"epoch": 0.2581318036633857,
"grad_norm": 0.32250335812568665,
"learning_rate": 0.0004674006675526214,
"loss": 0.9994050025939941,
"step": 990
},
{
"epoch": 0.26073919561958153,
"grad_norm": 0.3504716157913208,
"learning_rate": 0.0004665881168767346,
"loss": 1.0787917137145997,
"step": 1000
},
{
"epoch": 0.26073919561958153,
"eval_loss": 2.0505831241607666,
"eval_runtime": 60.8164,
"eval_samples_per_second": 9.734,
"eval_steps_per_second": 1.217,
"step": 1000
},
{
"epoch": 0.26334658757577734,
"grad_norm": 0.37507134675979614,
"learning_rate": 0.0004657662882867821,
"loss": 1.0458871841430664,
"step": 1010
},
{
"epoch": 0.26595397953197314,
"grad_norm": 0.34870800375938416,
"learning_rate": 0.00046493521698716536,
"loss": 1.0229036331176757,
"step": 1020
},
{
"epoch": 0.26856137148816894,
"grad_norm": 0.3507484197616577,
"learning_rate": 0.00046409493857821273,
"loss": 1.0416258811950683,
"step": 1030
},
{
"epoch": 0.27116876344436475,
"grad_norm": 0.28918078541755676,
"learning_rate": 0.0004632454890546547,
"loss": 0.974825668334961,
"step": 1040
},
{
"epoch": 0.2737761554005606,
"grad_norm": 0.29285478591918945,
"learning_rate": 0.0004623869048040817,
"loss": 0.9255588531494141,
"step": 1050
},
{
"epoch": 0.2763835473567564,
"grad_norm": 0.29449450969696045,
"learning_rate": 0.0004615192226053855,
"loss": 0.9860255241394043,
"step": 1060
},
{
"epoch": 0.2789909393129522,
"grad_norm": 0.3053012192249298,
"learning_rate": 0.0004606424796271834,
"loss": 0.9720993995666504,
"step": 1070
},
{
"epoch": 0.281598331269148,
"grad_norm": 0.2997720241546631,
"learning_rate": 0.0004597567134262266,
"loss": 0.9578587532043457,
"step": 1080
},
{
"epoch": 0.2842057232253439,
"grad_norm": 0.29879316687583923,
"learning_rate": 0.00045886196194579133,
"loss": 0.8845463752746582,
"step": 1090
},
{
"epoch": 0.2868131151815397,
"grad_norm": 0.32789888978004456,
"learning_rate": 0.00045795826351405276,
"loss": 1.0062164306640624,
"step": 1100
},
{
"epoch": 0.2894205071377355,
"grad_norm": 0.2998460829257965,
"learning_rate": 0.00045704565684244415,
"loss": 0.9499244689941406,
"step": 1110
},
{
"epoch": 0.2920278990939313,
"grad_norm": 0.34664222598075867,
"learning_rate": 0.00045612418102399785,
"loss": 0.937098503112793,
"step": 1120
},
{
"epoch": 0.2946352910501271,
"grad_norm": 0.34657037258148193,
"learning_rate": 0.000455193875531671,
"loss": 0.9578360557556153,
"step": 1130
},
{
"epoch": 0.29724268300632295,
"grad_norm": 0.3193321228027344,
"learning_rate": 0.0004542547802166546,
"loss": 0.9735245704650879,
"step": 1140
},
{
"epoch": 0.29985007496251875,
"grad_norm": 0.32926997542381287,
"learning_rate": 0.0004533069353066664,
"loss": 0.925290870666504,
"step": 1150
},
{
"epoch": 0.30245746691871456,
"grad_norm": 0.33400624990463257,
"learning_rate": 0.0004523503814042277,
"loss": 0.9294220924377441,
"step": 1160
},
{
"epoch": 0.30506485887491036,
"grad_norm": 0.326541930437088,
"learning_rate": 0.0004513851594849241,
"loss": 0.9727860450744629,
"step": 1170
},
{
"epoch": 0.30767225083110616,
"grad_norm": 0.26173487305641174,
"learning_rate": 0.0004504113108956499,
"loss": 1.0453700065612792,
"step": 1180
},
{
"epoch": 0.310279642787302,
"grad_norm": 0.325322687625885,
"learning_rate": 0.00044942887735283755,
"loss": 0.9772602081298828,
"step": 1190
},
{
"epoch": 0.3128870347434978,
"grad_norm": 0.2983860671520233,
"learning_rate": 0.00044843790094067026,
"loss": 1.0045940399169921,
"step": 1200
},
{
"epoch": 0.31549442669969363,
"grad_norm": 0.30153506994247437,
"learning_rate": 0.0004474384241092793,
"loss": 0.9639657020568848,
"step": 1210
},
{
"epoch": 0.31810181865588943,
"grad_norm": 0.272580087184906,
"learning_rate": 0.0004464304896729257,
"loss": 1.0098575592041015,
"step": 1220
},
{
"epoch": 0.32070921061208524,
"grad_norm": 0.311614066362381,
"learning_rate": 0.00044541414080816573,
"loss": 0.9908422470092774,
"step": 1230
},
{
"epoch": 0.3233166025682811,
"grad_norm": 0.33255434036254883,
"learning_rate": 0.0004443894210520024,
"loss": 1.0101188659667968,
"step": 1240
},
{
"epoch": 0.3259239945244769,
"grad_norm": 0.33257848024368286,
"learning_rate": 0.0004433563743000195,
"loss": 0.9614145278930664,
"step": 1250
},
{
"epoch": 0.3285313864806727,
"grad_norm": 0.37168046832084656,
"learning_rate": 0.00044231504480450145,
"loss": 0.9776069641113281,
"step": 1260
},
{
"epoch": 0.3311387784368685,
"grad_norm": 0.29584354162216187,
"learning_rate": 0.0004412654771725382,
"loss": 0.9117624282836914,
"step": 1270
},
{
"epoch": 0.3337461703930643,
"grad_norm": 0.36581864953041077,
"learning_rate": 0.0004402077163641137,
"loss": 0.9471235275268555,
"step": 1280
},
{
"epoch": 0.33635356234926017,
"grad_norm": 0.31348204612731934,
"learning_rate": 0.00043914180769018073,
"loss": 0.9160548210144043,
"step": 1290
},
{
"epoch": 0.338960954305456,
"grad_norm": 0.31987395882606506,
"learning_rate": 0.00043806779681071907,
"loss": 0.9445430755615234,
"step": 1300
},
{
"epoch": 0.3415683462616518,
"grad_norm": 0.2993371784687042,
"learning_rate": 0.00043698572973278026,
"loss": 0.926976203918457,
"step": 1310
},
{
"epoch": 0.3441757382178476,
"grad_norm": 0.29946768283843994,
"learning_rate": 0.0004358956528085165,
"loss": 0.9258660316467285,
"step": 1320
},
{
"epoch": 0.3467831301740434,
"grad_norm": 0.31041479110717773,
"learning_rate": 0.0004347976127331953,
"loss": 1.0128493309020996,
"step": 1330
},
{
"epoch": 0.34939052213023924,
"grad_norm": 0.2935703992843628,
"learning_rate": 0.0004336916565431987,
"loss": 0.9315634727478027,
"step": 1340
},
{
"epoch": 0.35199791408643505,
"grad_norm": 0.3416391909122467,
"learning_rate": 0.00043257783161400917,
"loss": 0.988331413269043,
"step": 1350
},
{
"epoch": 0.35460530604263085,
"grad_norm": 0.31745293736457825,
"learning_rate": 0.00043145618565817946,
"loss": 0.9010594367980957,
"step": 1360
},
{
"epoch": 0.35721269799882666,
"grad_norm": 0.34742966294288635,
"learning_rate": 0.00043032676672328916,
"loss": 0.9530973434448242,
"step": 1370
},
{
"epoch": 0.3598200899550225,
"grad_norm": 0.29915374517440796,
"learning_rate": 0.00042918962318988664,
"loss": 0.9269295692443847,
"step": 1380
},
{
"epoch": 0.3624274819112183,
"grad_norm": 0.29669490456581116,
"learning_rate": 0.00042804480376941597,
"loss": 0.9117980003356934,
"step": 1390
},
{
"epoch": 0.3650348738674141,
"grad_norm": 0.3245804011821747,
"learning_rate": 0.00042689235750213093,
"loss": 0.9205721855163574,
"step": 1400
},
{
"epoch": 0.3676422658236099,
"grad_norm": 0.30812278389930725,
"learning_rate": 0.000425732333754994,
"loss": 0.9146486282348633,
"step": 1410
},
{
"epoch": 0.37024965777980573,
"grad_norm": 0.3339979648590088,
"learning_rate": 0.0004245647822195616,
"loss": 0.8891145706176757,
"step": 1420
},
{
"epoch": 0.3728570497360016,
"grad_norm": 0.33809196949005127,
"learning_rate": 0.0004233897529098556,
"loss": 1.0212121963500977,
"step": 1430
},
{
"epoch": 0.3754644416921974,
"grad_norm": 0.2943659722805023,
"learning_rate": 0.0004222072961602209,
"loss": 0.914692497253418,
"step": 1440
},
{
"epoch": 0.3780718336483932,
"grad_norm": 0.3399428427219391,
"learning_rate": 0.000421017462623169,
"loss": 0.9326979637145996,
"step": 1450
},
{
"epoch": 0.380679225604589,
"grad_norm": 0.32993045449256897,
"learning_rate": 0.00041982030326720866,
"loss": 0.8850909233093261,
"step": 1460
},
{
"epoch": 0.3832866175607848,
"grad_norm": 0.337248831987381,
"learning_rate": 0.0004186158693746622,
"loss": 0.9511479377746582,
"step": 1470
},
{
"epoch": 0.38589400951698066,
"grad_norm": 0.3178328275680542,
"learning_rate": 0.0004174042125394689,
"loss": 0.8905654907226562,
"step": 1480
},
{
"epoch": 0.38850140147317647,
"grad_norm": 0.2901822328567505,
"learning_rate": 0.00041618538466497496,
"loss": 0.8919774055480957,
"step": 1490
},
{
"epoch": 0.39110879342937227,
"grad_norm": 0.31942248344421387,
"learning_rate": 0.0004149594379617099,
"loss": 0.9066588401794433,
"step": 1500
},
{
"epoch": 0.39110879342937227,
"eval_loss": 2.0470354557037354,
"eval_runtime": 60.8827,
"eval_samples_per_second": 9.724,
"eval_steps_per_second": 1.215,
"step": 1500
},
{
"epoch": 0.3937161853855681,
"grad_norm": 0.30218419432640076,
"learning_rate": 0.0004137264249451501,
"loss": 0.8638726234436035,
"step": 1510
},
{
"epoch": 0.3963235773417639,
"grad_norm": 0.41076040267944336,
"learning_rate": 0.00041248639843346953,
"loss": 0.9257563591003418,
"step": 1520
},
{
"epoch": 0.39893096929795974,
"grad_norm": 0.3387877643108368,
"learning_rate": 0.0004112394115452768,
"loss": 0.923636531829834,
"step": 1530
},
{
"epoch": 0.40153836125415554,
"grad_norm": 0.293745756149292,
"learning_rate": 0.0004099855176973399,
"loss": 0.8279055595397949,
"step": 1540
},
{
"epoch": 0.40414575321035134,
"grad_norm": 0.2916512191295624,
"learning_rate": 0.00040872477060229797,
"loss": 0.8548236846923828,
"step": 1550
},
{
"epoch": 0.40675314516654715,
"grad_norm": 0.33279502391815186,
"learning_rate": 0.00040745722426636043,
"loss": 0.8940540313720703,
"step": 1560
},
{
"epoch": 0.40936053712274295,
"grad_norm": 0.3562711477279663,
"learning_rate": 0.0004061829329869937,
"loss": 0.8976078987121582,
"step": 1570
},
{
"epoch": 0.4119679290789388,
"grad_norm": 0.3408759832382202,
"learning_rate": 0.00040490195135059503,
"loss": 0.943824291229248,
"step": 1580
},
{
"epoch": 0.4145753210351346,
"grad_norm": 0.2810667157173157,
"learning_rate": 0.0004036143342301542,
"loss": 0.915517520904541,
"step": 1590
},
{
"epoch": 0.4171827129913304,
"grad_norm": 0.43379518389701843,
"learning_rate": 0.00040232013678290316,
"loss": 0.9169961929321289,
"step": 1600
},
{
"epoch": 0.4197901049475262,
"grad_norm": 0.3522005081176758,
"learning_rate": 0.000401019414447953,
"loss": 0.8237640380859375,
"step": 1610
},
{
"epoch": 0.422397496903722,
"grad_norm": 0.2890526354312897,
"learning_rate": 0.0003997122229439196,
"loss": 0.8344717025756836,
"step": 1620
},
{
"epoch": 0.4250048888599179,
"grad_norm": 0.2849102020263672,
"learning_rate": 0.0003983986182665362,
"loss": 0.8765329360961914,
"step": 1630
},
{
"epoch": 0.4276122808161137,
"grad_norm": 0.26151859760284424,
"learning_rate": 0.0003970786566862553,
"loss": 0.8577868461608886,
"step": 1640
},
{
"epoch": 0.4302196727723095,
"grad_norm": 0.44217562675476074,
"learning_rate": 0.0003957523947458377,
"loss": 0.9197720527648926,
"step": 1650
},
{
"epoch": 0.4328270647285053,
"grad_norm": 0.318926066160202,
"learning_rate": 0.0003944198892579309,
"loss": 0.9023479461669922,
"step": 1660
},
{
"epoch": 0.43543445668470115,
"grad_norm": 0.3373873233795166,
"learning_rate": 0.00039308119730263494,
"loss": 0.8540445327758789,
"step": 1670
},
{
"epoch": 0.43804184864089696,
"grad_norm": 0.3525305688381195,
"learning_rate": 0.0003917363762250576,
"loss": 0.9312978744506836,
"step": 1680
},
{
"epoch": 0.44064924059709276,
"grad_norm": 0.32791227102279663,
"learning_rate": 0.00039038548363285825,
"loss": 0.9128911018371582,
"step": 1690
},
{
"epoch": 0.44325663255328857,
"grad_norm": 0.3262438476085663,
"learning_rate": 0.0003890285773937792,
"loss": 0.8420479774475098,
"step": 1700
},
{
"epoch": 0.44586402450948437,
"grad_norm": 0.3806127607822418,
"learning_rate": 0.00038766571563316756,
"loss": 0.9506552696228028,
"step": 1710
},
{
"epoch": 0.44847141646568023,
"grad_norm": 0.35995638370513916,
"learning_rate": 0.0003862969567314852,
"loss": 0.881410026550293,
"step": 1720
},
{
"epoch": 0.45107880842187603,
"grad_norm": 0.34164267778396606,
"learning_rate": 0.0003849223593218082,
"loss": 0.8768704414367676,
"step": 1730
},
{
"epoch": 0.45368620037807184,
"grad_norm": 0.27599456906318665,
"learning_rate": 0.00038354198228731414,
"loss": 0.7965556621551514,
"step": 1740
},
{
"epoch": 0.45629359233426764,
"grad_norm": 0.32938748598098755,
"learning_rate": 0.00038215588475876117,
"loss": 0.9088653564453125,
"step": 1750
},
{
"epoch": 0.45890098429046344,
"grad_norm": 0.3628661036491394,
"learning_rate": 0.000380764126111954,
"loss": 0.8792169570922852,
"step": 1760
},
{
"epoch": 0.4615083762466593,
"grad_norm": 0.32461732625961304,
"learning_rate": 0.0003793667659652011,
"loss": 0.8721429824829101,
"step": 1770
},
{
"epoch": 0.4641157682028551,
"grad_norm": 0.29975461959838867,
"learning_rate": 0.0003779638641767602,
"loss": 0.9021189689636231,
"step": 1780
},
{
"epoch": 0.4667231601590509,
"grad_norm": 0.3076811134815216,
"learning_rate": 0.00037655548084227484,
"loss": 0.9223040580749512,
"step": 1790
},
{
"epoch": 0.4693305521152467,
"grad_norm": 0.2991771101951599,
"learning_rate": 0.00037514167629219955,
"loss": 0.8949285507202148,
"step": 1800
},
{
"epoch": 0.4719379440714425,
"grad_norm": 0.3248666226863861,
"learning_rate": 0.0003737225110892156,
"loss": 0.875214958190918,
"step": 1810
},
{
"epoch": 0.4745453360276384,
"grad_norm": 0.2874812185764313,
"learning_rate": 0.0003722980460256372,
"loss": 0.8205836296081543,
"step": 1820
},
{
"epoch": 0.4771527279838342,
"grad_norm": 0.3912108540534973,
"learning_rate": 0.0003708683421208066,
"loss": 0.9452132225036621,
"step": 1830
},
{
"epoch": 0.47976011994003,
"grad_norm": 0.35194849967956543,
"learning_rate": 0.00036943346061848054,
"loss": 0.8853225708007812,
"step": 1840
},
{
"epoch": 0.4823675118962258,
"grad_norm": 0.2641483545303345,
"learning_rate": 0.000367993462984207,
"loss": 0.8014869689941406,
"step": 1850
},
{
"epoch": 0.4849749038524216,
"grad_norm": 0.2862023711204529,
"learning_rate": 0.000366548410902692,
"loss": 0.7931074619293212,
"step": 1860
},
{
"epoch": 0.48758229580861745,
"grad_norm": 0.3328746259212494,
"learning_rate": 0.0003650983662751571,
"loss": 0.8649985313415527,
"step": 1870
},
{
"epoch": 0.49018968776481325,
"grad_norm": 0.33828577399253845,
"learning_rate": 0.0003636433912166881,
"loss": 0.8727825164794922,
"step": 1880
},
{
"epoch": 0.49279707972100906,
"grad_norm": 0.3215215802192688,
"learning_rate": 0.0003621835480535739,
"loss": 0.7987274169921875,
"step": 1890
},
{
"epoch": 0.49540447167720486,
"grad_norm": 0.2811441421508789,
"learning_rate": 0.00036071889932063706,
"loss": 0.761207914352417,
"step": 1900
},
{
"epoch": 0.49801186363340066,
"grad_norm": 0.28584030270576477,
"learning_rate": 0.0003592495077585543,
"loss": 0.8611610412597657,
"step": 1910
},
{
"epoch": 0.5006192555895965,
"grad_norm": 0.3160797357559204,
"learning_rate": 0.00035777543631116977,
"loss": 0.887575340270996,
"step": 1920
},
{
"epoch": 0.5032266475457923,
"grad_norm": 0.33388271927833557,
"learning_rate": 0.0003562967481227982,
"loss": 0.8079765319824219,
"step": 1930
},
{
"epoch": 0.5058340395019881,
"grad_norm": 0.2789759337902069,
"learning_rate": 0.0003548135065355201,
"loss": 0.7524425029754639,
"step": 1940
},
{
"epoch": 0.508441431458184,
"grad_norm": 0.3001299500465393,
"learning_rate": 0.00035332577508646846,
"loss": 0.8838024139404297,
"step": 1950
},
{
"epoch": 0.5110488234143797,
"grad_norm": 0.3202671706676483,
"learning_rate": 0.000351833617505107,
"loss": 0.8436161994934082,
"step": 1960
},
{
"epoch": 0.5136562153705756,
"grad_norm": 0.30766424536705017,
"learning_rate": 0.0003503370977105002,
"loss": 0.8779410362243653,
"step": 1970
},
{
"epoch": 0.5162636073267713,
"grad_norm": 0.3019047975540161,
"learning_rate": 0.00034883627980857523,
"loss": 0.7935092449188232,
"step": 1980
},
{
"epoch": 0.5188709992829672,
"grad_norm": 0.31746894121170044,
"learning_rate": 0.000347331228089376,
"loss": 0.7767213821411133,
"step": 1990
},
{
"epoch": 0.5214783912391631,
"grad_norm": 0.32700610160827637,
"learning_rate": 0.0003458220070243089,
"loss": 0.8765421867370605,
"step": 2000
},
{
"epoch": 0.5214783912391631,
"eval_loss": 2.087125301361084,
"eval_runtime": 60.8303,
"eval_samples_per_second": 9.732,
"eval_steps_per_second": 1.216,
"step": 2000
},
{
"epoch": 0.5240857831953588,
"grad_norm": 0.3013451397418976,
"learning_rate": 0.0003443086812633812,
"loss": 0.7956989765167236,
"step": 2010
},
{
"epoch": 0.5266931751515547,
"grad_norm": 0.32685521245002747,
"learning_rate": 0.0003427913156324319,
"loss": 0.8650039672851563,
"step": 2020
},
{
"epoch": 0.5293005671077504,
"grad_norm": 0.29438385367393494,
"learning_rate": 0.0003412699751303544,
"loss": 0.8095410346984864,
"step": 2030
},
{
"epoch": 0.5319079590639463,
"grad_norm": 0.29924049973487854,
"learning_rate": 0.00033974472492631234,
"loss": 0.8555173873901367,
"step": 2040
},
{
"epoch": 0.5345153510201421,
"grad_norm": 0.290536105632782,
"learning_rate": 0.0003382156303569481,
"loss": 0.8646109580993653,
"step": 2050
},
{
"epoch": 0.5371227429763379,
"grad_norm": 0.35994812846183777,
"learning_rate": 0.0003366827569235838,
"loss": 0.8643518447875976,
"step": 2060
},
{
"epoch": 0.5397301349325337,
"grad_norm": 0.36780649423599243,
"learning_rate": 0.00033514617028941547,
"loss": 0.8244940757751464,
"step": 2070
},
{
"epoch": 0.5423375268887295,
"grad_norm": 0.37980470061302185,
"learning_rate": 0.0003336059362767004,
"loss": 0.8342851638793946,
"step": 2080
},
{
"epoch": 0.5449449188449254,
"grad_norm": 0.3380134105682373,
"learning_rate": 0.0003320621208639374,
"loss": 0.8450146675109863,
"step": 2090
},
{
"epoch": 0.5475523108011212,
"grad_norm": 0.35600146651268005,
"learning_rate": 0.00033051479018304054,
"loss": 0.8487363815307617,
"step": 2100
},
{
"epoch": 0.550159702757317,
"grad_norm": 0.3214777708053589,
"learning_rate": 0.0003289640105165063,
"loss": 0.8061514854431152,
"step": 2110
},
{
"epoch": 0.5527670947135128,
"grad_norm": 0.3083333671092987,
"learning_rate": 0.00032740984829457404,
"loss": 0.7891229152679443,
"step": 2120
},
{
"epoch": 0.5553744866697087,
"grad_norm": 0.29040274024009705,
"learning_rate": 0.000325852370092381,
"loss": 0.7813839435577392,
"step": 2130
},
{
"epoch": 0.5579818786259044,
"grad_norm": 0.3533709645271301,
"learning_rate": 0.00032429164262710934,
"loss": 0.8251482009887695,
"step": 2140
},
{
"epoch": 0.5605892705821003,
"grad_norm": 0.3043120205402374,
"learning_rate": 0.00032272773275512933,
"loss": 0.8363678932189942,
"step": 2150
},
{
"epoch": 0.563196662538296,
"grad_norm": 0.3815644085407257,
"learning_rate": 0.00032116070746913484,
"loss": 0.8239590644836425,
"step": 2160
},
{
"epoch": 0.5658040544944919,
"grad_norm": 0.3189505934715271,
"learning_rate": 0.0003195906338952736,
"loss": 0.7820507526397705,
"step": 2170
},
{
"epoch": 0.5684114464506878,
"grad_norm": 0.3006654679775238,
"learning_rate": 0.00031801757929027187,
"loss": 0.7826581954956054,
"step": 2180
},
{
"epoch": 0.5710188384068835,
"grad_norm": 0.35971957445144653,
"learning_rate": 0.0003164416110385534,
"loss": 0.8313183784484863,
"step": 2190
},
{
"epoch": 0.5736262303630794,
"grad_norm": 0.35061341524124146,
"learning_rate": 0.000314862796649353,
"loss": 0.7722214221954345,
"step": 2200
},
{
"epoch": 0.5762336223192751,
"grad_norm": 0.28702011704444885,
"learning_rate": 0.00031328120375382414,
"loss": 0.7329557418823243,
"step": 2210
},
{
"epoch": 0.578841014275471,
"grad_norm": 0.3457954525947571,
"learning_rate": 0.0003116969001021427,
"loss": 0.7929254531860351,
"step": 2220
},
{
"epoch": 0.5814484062316668,
"grad_norm": 0.3249218463897705,
"learning_rate": 0.00031010995356060416,
"loss": 0.7590729236602783,
"step": 2230
},
{
"epoch": 0.5840557981878626,
"grad_norm": 0.26487141847610474,
"learning_rate": 0.0003085204321087165,
"loss": 0.7542837142944336,
"step": 2240
},
{
"epoch": 0.5866631901440584,
"grad_norm": 0.33252763748168945,
"learning_rate": 0.00030692840383628845,
"loss": 0.8257474899291992,
"step": 2250
},
{
"epoch": 0.5892705821002542,
"grad_norm": 0.3328820765018463,
"learning_rate": 0.00030533393694051256,
"loss": 0.8061025619506836,
"step": 2260
},
{
"epoch": 0.59187797405645,
"grad_norm": 0.292125403881073,
"learning_rate": 0.0003037370997230436,
"loss": 0.7719024181365967,
"step": 2270
},
{
"epoch": 0.5944853660126459,
"grad_norm": 0.330931156873703,
"learning_rate": 0.0003021379605870736,
"loss": 0.7803617000579834,
"step": 2280
},
{
"epoch": 0.5970927579688416,
"grad_norm": 0.3659128248691559,
"learning_rate": 0.00030053658803440064,
"loss": 0.8258522987365723,
"step": 2290
},
{
"epoch": 0.5997001499250375,
"grad_norm": 0.3726382255554199,
"learning_rate": 0.0002989330506624951,
"loss": 0.8412979125976563,
"step": 2300
},
{
"epoch": 0.6023075418812333,
"grad_norm": 0.3532240092754364,
"learning_rate": 0.000297327417161561,
"loss": 0.7726183891296386,
"step": 2310
},
{
"epoch": 0.6049149338374291,
"grad_norm": 0.3252621591091156,
"learning_rate": 0.00029571975631159355,
"loss": 0.775645637512207,
"step": 2320
},
{
"epoch": 0.607522325793625,
"grad_norm": 0.28014039993286133,
"learning_rate": 0.00029411013697943294,
"loss": 0.7973006248474122,
"step": 2330
},
{
"epoch": 0.6101297177498207,
"grad_norm": 0.3317548930644989,
"learning_rate": 0.000292498628115814,
"loss": 0.7408670425415039,
"step": 2340
},
{
"epoch": 0.6127371097060166,
"grad_norm": 0.3302936851978302,
"learning_rate": 0.00029088529875241306,
"loss": 0.7508449554443359,
"step": 2350
},
{
"epoch": 0.6153445016622123,
"grad_norm": 0.3802710473537445,
"learning_rate": 0.0002892702179988903,
"loss": 0.7675702571868896,
"step": 2360
},
{
"epoch": 0.6179518936184082,
"grad_norm": 0.31116169691085815,
"learning_rate": 0.00028765345503993,
"loss": 0.7534912109375,
"step": 2370
},
{
"epoch": 0.620559285574604,
"grad_norm": 0.3058822453022003,
"learning_rate": 0.00028603507913227617,
"loss": 0.7343992233276367,
"step": 2380
},
{
"epoch": 0.6231666775307998,
"grad_norm": 0.35350584983825684,
"learning_rate": 0.0002844151596017665,
"loss": 0.7853089332580566,
"step": 2390
},
{
"epoch": 0.6257740694869957,
"grad_norm": 0.30020278692245483,
"learning_rate": 0.00028279376584036187,
"loss": 0.763953161239624,
"step": 2400
},
{
"epoch": 0.6283814614431914,
"grad_norm": 0.3629872798919678,
"learning_rate": 0.00028117096730317475,
"loss": 0.7954237937927247,
"step": 2410
},
{
"epoch": 0.6309888533993873,
"grad_norm": 0.25756698846817017,
"learning_rate": 0.0002795468335054935,
"loss": 0.7064559936523438,
"step": 2420
},
{
"epoch": 0.6335962453555831,
"grad_norm": 0.36921486258506775,
"learning_rate": 0.00027792143401980435,
"loss": 0.7957570552825928,
"step": 2430
},
{
"epoch": 0.6362036373117789,
"grad_norm": 0.37634629011154175,
"learning_rate": 0.0002762948384728115,
"loss": 0.7668004035949707,
"step": 2440
},
{
"epoch": 0.6388110292679747,
"grad_norm": 0.33615532517433167,
"learning_rate": 0.0002746671165424545,
"loss": 0.7280832290649414,
"step": 2450
},
{
"epoch": 0.6414184212241705,
"grad_norm": 0.33129194378852844,
"learning_rate": 0.0002730383379549234,
"loss": 0.735734510421753,
"step": 2460
},
{
"epoch": 0.6440258131803663,
"grad_norm": 0.3507666289806366,
"learning_rate": 0.0002714085724816718,
"loss": 0.7372643947601318,
"step": 2470
},
{
"epoch": 0.6466332051365622,
"grad_norm": 0.347154825925827,
"learning_rate": 0.0002697778899364284,
"loss": 0.7815746784210205,
"step": 2480
},
{
"epoch": 0.6492405970927579,
"grad_norm": 0.2843272089958191,
"learning_rate": 0.000268146360172206,
"loss": 0.7554127693176269,
"step": 2490
},
{
"epoch": 0.6518479890489538,
"grad_norm": 0.3696630001068115,
"learning_rate": 0.0002665140530783097,
"loss": 0.7425910472869873,
"step": 2500
},
{
"epoch": 0.6518479890489538,
"eval_loss": 2.13978910446167,
"eval_runtime": 60.7803,
"eval_samples_per_second": 9.74,
"eval_steps_per_second": 1.217,
"step": 2500
},
{
"epoch": 0.6544553810051496,
"grad_norm": 0.3483107089996338,
"learning_rate": 0.0002648810385773424,
"loss": 0.7357473373413086,
"step": 2510
},
{
"epoch": 0.6570627729613454,
"grad_norm": 0.35009869933128357,
"learning_rate": 0.0002632473866222104,
"loss": 0.790524959564209,
"step": 2520
},
{
"epoch": 0.6596701649175413,
"grad_norm": 0.28543636202812195,
"learning_rate": 0.00026161316719312637,
"loss": 0.7308042049407959,
"step": 2530
},
{
"epoch": 0.662277556873737,
"grad_norm": 0.36112868785858154,
"learning_rate": 0.00025997845029461134,
"loss": 0.7709194183349609,
"step": 2540
},
{
"epoch": 0.6648849488299329,
"grad_norm": 0.33462971448898315,
"learning_rate": 0.00025834330595249654,
"loss": 0.7110116004943847,
"step": 2550
},
{
"epoch": 0.6674923407861286,
"grad_norm": 0.29224660992622375,
"learning_rate": 0.0002567078042109235,
"loss": 0.7173576831817627,
"step": 2560
},
{
"epoch": 0.6700997327423245,
"grad_norm": 0.3267500400543213,
"learning_rate": 0.00025507201512934327,
"loss": 0.7058556079864502,
"step": 2570
},
{
"epoch": 0.6727071246985203,
"grad_norm": 0.3215857148170471,
"learning_rate": 0.0002534360087795159,
"loss": 0.7735134124755859,
"step": 2580
},
{
"epoch": 0.6753145166547161,
"grad_norm": 0.29826200008392334,
"learning_rate": 0.0002517998552425083,
"loss": 0.6976490020751953,
"step": 2590
},
{
"epoch": 0.677921908610912,
"grad_norm": 0.3853761553764343,
"learning_rate": 0.00025016362460569255,
"loss": 0.7154698848724366,
"step": 2600
},
{
"epoch": 0.6805293005671077,
"grad_norm": 0.32427939772605896,
"learning_rate": 0.0002485273869597433,
"loss": 0.7396796226501465,
"step": 2610
},
{
"epoch": 0.6831366925233036,
"grad_norm": 0.28578534722328186,
"learning_rate": 0.0002468912123956354,
"loss": 0.7164602756500245,
"step": 2620
},
{
"epoch": 0.6857440844794994,
"grad_norm": 0.36031123995780945,
"learning_rate": 0.00024525517100164166,
"loss": 0.767027473449707,
"step": 2630
},
{
"epoch": 0.6883514764356952,
"grad_norm": 0.3250391483306885,
"learning_rate": 0.0002436193328603301,
"loss": 0.7324582576751709,
"step": 2640
},
{
"epoch": 0.690958868391891,
"grad_norm": 0.3187474012374878,
"learning_rate": 0.00024198376804556235,
"loss": 0.6928200721740723,
"step": 2650
},
{
"epoch": 0.6935662603480868,
"grad_norm": 0.3236410319805145,
"learning_rate": 0.00024034854661949152,
"loss": 0.7143116474151612,
"step": 2660
},
{
"epoch": 0.6961736523042826,
"grad_norm": 0.32728391885757446,
"learning_rate": 0.00023871373862956088,
"loss": 0.7109772205352783,
"step": 2670
},
{
"epoch": 0.6987810442604785,
"grad_norm": 0.2942025661468506,
"learning_rate": 0.0002370794141055038,
"loss": 0.7113640785217286,
"step": 2680
},
{
"epoch": 0.7013884362166742,
"grad_norm": 0.34320199489593506,
"learning_rate": 0.00023544564305634342,
"loss": 0.7355948448181152,
"step": 2690
},
{
"epoch": 0.7039958281728701,
"grad_norm": 0.2952670753002167,
"learning_rate": 0.0002338124954673938,
"loss": 0.7053794860839844,
"step": 2700
},
{
"epoch": 0.706603220129066,
"grad_norm": 0.34913331270217896,
"learning_rate": 0.00023218004129726218,
"loss": 0.7259132862091064,
"step": 2710
},
{
"epoch": 0.7092106120852617,
"grad_norm": 0.29017773270606995,
"learning_rate": 0.00023054835047485193,
"loss": 0.6789509773254394,
"step": 2720
},
{
"epoch": 0.7118180040414576,
"grad_norm": 0.344415545463562,
"learning_rate": 0.00022891749289636725,
"loss": 0.6990691661834717,
"step": 2730
},
{
"epoch": 0.7144253959976533,
"grad_norm": 0.3235209584236145,
"learning_rate": 0.00022728753842231886,
"loss": 0.7382527351379394,
"step": 2740
},
{
"epoch": 0.7170327879538492,
"grad_norm": 0.2997324466705322,
"learning_rate": 0.0002256585568745315,
"loss": 0.711235761642456,
"step": 2750
},
{
"epoch": 0.719640179910045,
"grad_norm": 0.33747562766075134,
"learning_rate": 0.000224030618033153,
"loss": 0.7032158374786377,
"step": 2760
},
{
"epoch": 0.7222475718662408,
"grad_norm": 0.34918609261512756,
"learning_rate": 0.00022240379163366523,
"loss": 0.6888625144958496,
"step": 2770
},
{
"epoch": 0.7248549638224366,
"grad_norm": 0.32571062445640564,
"learning_rate": 0.00022077814736389648,
"loss": 0.6955374717712403,
"step": 2780
},
{
"epoch": 0.7274623557786324,
"grad_norm": 0.36113062500953674,
"learning_rate": 0.00021915375486103675,
"loss": 0.7120633602142334,
"step": 2790
},
{
"epoch": 0.7300697477348282,
"grad_norm": 0.32838910818099976,
"learning_rate": 0.00021753068370865454,
"loss": 0.6790756225585938,
"step": 2800
},
{
"epoch": 0.7326771396910241,
"grad_norm": 0.29126304388046265,
"learning_rate": 0.00021590900343371598,
"loss": 0.646269702911377,
"step": 2810
},
{
"epoch": 0.7352845316472199,
"grad_norm": 0.3991219997406006,
"learning_rate": 0.00021428878350360667,
"loss": 0.6770219802856445,
"step": 2820
},
{
"epoch": 0.7378919236034157,
"grad_norm": 0.2936253249645233,
"learning_rate": 0.000212670093323156,
"loss": 0.6455300331115723,
"step": 2830
},
{
"epoch": 0.7404993155596115,
"grad_norm": 0.3143956661224365,
"learning_rate": 0.0002110530022316637,
"loss": 0.6527771949768066,
"step": 2840
},
{
"epoch": 0.7431067075158073,
"grad_norm": 0.36326706409454346,
"learning_rate": 0.0002094375794999301,
"loss": 0.6709997177124023,
"step": 2850
},
{
"epoch": 0.7457140994720032,
"grad_norm": 0.3787069022655487,
"learning_rate": 0.0002078238943272886,
"loss": 0.6706299781799316,
"step": 2860
},
{
"epoch": 0.7483214914281989,
"grad_norm": 0.3385184705257416,
"learning_rate": 0.00020621201583864108,
"loss": 0.6603190422058105,
"step": 2870
},
{
"epoch": 0.7509288833843948,
"grad_norm": 0.3112833797931671,
"learning_rate": 0.00020460201308149727,
"loss": 0.6779971122741699,
"step": 2880
},
{
"epoch": 0.7535362753405905,
"grad_norm": 0.365662157535553,
"learning_rate": 0.00020299395502301689,
"loss": 0.7141220092773437,
"step": 2890
},
{
"epoch": 0.7561436672967864,
"grad_norm": 0.32723233103752136,
"learning_rate": 0.00020138791054705505,
"loss": 0.6945212364196778,
"step": 2900
},
{
"epoch": 0.7587510592529823,
"grad_norm": 0.29125431180000305,
"learning_rate": 0.00019978394845121173,
"loss": 0.6242878437042236,
"step": 2910
},
{
"epoch": 0.761358451209178,
"grad_norm": 0.30191484093666077,
"learning_rate": 0.0001981821374438848,
"loss": 0.6447752952575684,
"step": 2920
},
{
"epoch": 0.7639658431653739,
"grad_norm": 0.2737407386302948,
"learning_rate": 0.0001965825461413265,
"loss": 0.6486302852630615,
"step": 2930
},
{
"epoch": 0.7665732351215696,
"grad_norm": 0.3413922190666199,
"learning_rate": 0.00019498524306470438,
"loss": 0.6337445259094239,
"step": 2940
},
{
"epoch": 0.7691806270777655,
"grad_norm": 0.3383452594280243,
"learning_rate": 0.00019339029663716613,
"loss": 0.7187496185302734,
"step": 2950
},
{
"epoch": 0.7717880190339613,
"grad_norm": 0.3658270835876465,
"learning_rate": 0.0001917977751809082,
"loss": 0.6551912784576416,
"step": 2960
},
{
"epoch": 0.7743954109901571,
"grad_norm": 0.3562202751636505,
"learning_rate": 0.00019020774691424946,
"loss": 0.6773749351501465,
"step": 2970
},
{
"epoch": 0.7770028029463529,
"grad_norm": 0.3962496817111969,
"learning_rate": 0.00018862027994870887,
"loss": 0.7006560325622558,
"step": 2980
},
{
"epoch": 0.7796101949025487,
"grad_norm": 0.27267396450042725,
"learning_rate": 0.00018703544228608775,
"loss": 0.6216824054718018,
"step": 2990
},
{
"epoch": 0.7822175868587445,
"grad_norm": 0.3390612304210663,
"learning_rate": 0.0001854533018155567,
"loss": 0.6901330471038818,
"step": 3000
},
{
"epoch": 0.7822175868587445,
"eval_loss": 2.226897716522217,
"eval_runtime": 60.8078,
"eval_samples_per_second": 9.736,
"eval_steps_per_second": 1.217,
"step": 3000
},
{
"epoch": 0.7848249788149404,
"grad_norm": 0.3287055790424347,
"learning_rate": 0.00018387392631074778,
"loss": 0.7144739627838135,
"step": 3010
},
{
"epoch": 0.7874323707711361,
"grad_norm": 0.3081079423427582,
"learning_rate": 0.00018229738342685105,
"loss": 0.6702582836151123,
"step": 3020
},
{
"epoch": 0.790039762727332,
"grad_norm": 0.3318571448326111,
"learning_rate": 0.00018072374069771658,
"loss": 0.6706883907318115,
"step": 3030
},
{
"epoch": 0.7926471546835278,
"grad_norm": 0.3265221118927002,
"learning_rate": 0.00017915306553296128,
"loss": 0.6843328952789307,
"step": 3040
},
{
"epoch": 0.7952545466397236,
"grad_norm": 0.27763375639915466,
"learning_rate": 0.00017758542521508164,
"loss": 0.6655256748199463,
"step": 3050
},
{
"epoch": 0.7978619385959195,
"grad_norm": 0.37208816409111023,
"learning_rate": 0.00017602088689657136,
"loss": 0.6352975845336915,
"step": 3060
},
{
"epoch": 0.8004693305521152,
"grad_norm": 0.34267693758010864,
"learning_rate": 0.00017445951759704486,
"loss": 0.6490068912506104,
"step": 3070
},
{
"epoch": 0.8030767225083111,
"grad_norm": 0.3621560037136078,
"learning_rate": 0.00017290138420036623,
"loss": 0.6506584644317627,
"step": 3080
},
{
"epoch": 0.8056841144645068,
"grad_norm": 0.38878488540649414,
"learning_rate": 0.0001713465534517843,
"loss": 0.6434064388275147,
"step": 3090
},
{
"epoch": 0.8082915064207027,
"grad_norm": 0.3283291161060333,
"learning_rate": 0.00016979509195507353,
"loss": 0.652371883392334,
"step": 3100
},
{
"epoch": 0.8108988983768985,
"grad_norm": 0.3809128403663635,
"learning_rate": 0.00016824706616968074,
"loss": 0.6325668334960938,
"step": 3110
},
{
"epoch": 0.8135062903330943,
"grad_norm": 0.3271070718765259,
"learning_rate": 0.00016670254240787834,
"loss": 0.632191514968872,
"step": 3120
},
{
"epoch": 0.8161136822892902,
"grad_norm": 0.3322986662387848,
"learning_rate": 0.00016516158683192383,
"loss": 0.6598535060882569,
"step": 3130
},
{
"epoch": 0.8187210742454859,
"grad_norm": 0.31290119886398315,
"learning_rate": 0.00016362426545122517,
"loss": 0.6589827060699462,
"step": 3140
},
{
"epoch": 0.8213284662016818,
"grad_norm": 0.35526585578918457,
"learning_rate": 0.00016209064411951384,
"loss": 0.6144824028015137,
"step": 3150
},
{
"epoch": 0.8239358581578776,
"grad_norm": 0.317221462726593,
"learning_rate": 0.00016056078853202344,
"loss": 0.6057178497314453,
"step": 3160
},
{
"epoch": 0.8265432501140734,
"grad_norm": 0.35776281356811523,
"learning_rate": 0.00015903476422267547,
"loss": 0.6184150695800781,
"step": 3170
},
{
"epoch": 0.8291506420702692,
"grad_norm": 0.32412266731262207,
"learning_rate": 0.00015751263656127234,
"loss": 0.604495620727539,
"step": 3180
},
{
"epoch": 0.831758034026465,
"grad_norm": 0.3678344488143921,
"learning_rate": 0.00015599447075069714,
"loss": 0.6431977272033691,
"step": 3190
},
{
"epoch": 0.8343654259826608,
"grad_norm": 0.3430567681789398,
"learning_rate": 0.0001544803318241203,
"loss": 0.600965166091919,
"step": 3200
},
{
"epoch": 0.8369728179388567,
"grad_norm": 0.3765794038772583,
"learning_rate": 0.00015297028464221395,
"loss": 0.629162073135376,
"step": 3210
},
{
"epoch": 0.8395802098950524,
"grad_norm": 0.290073961019516,
"learning_rate": 0.00015146439389037355,
"loss": 0.6020309448242187,
"step": 3220
},
{
"epoch": 0.8421876018512483,
"grad_norm": 0.27276337146759033,
"learning_rate": 0.00014996272407594694,
"loss": 0.6155780792236328,
"step": 3230
},
{
"epoch": 0.844794993807444,
"grad_norm": 0.34680280089378357,
"learning_rate": 0.00014846533952547094,
"loss": 0.6324656486511231,
"step": 3240
},
{
"epoch": 0.8474023857636399,
"grad_norm": 0.358701229095459,
"learning_rate": 0.00014697230438191617,
"loss": 0.6087419033050537,
"step": 3250
},
{
"epoch": 0.8500097777198358,
"grad_norm": 0.3267561197280884,
"learning_rate": 0.00014548368260193888,
"loss": 0.6123754024505615,
"step": 3260
},
{
"epoch": 0.8526171696760315,
"grad_norm": 0.33160290122032166,
"learning_rate": 0.0001439995379531416,
"loss": 0.6234821796417236,
"step": 3270
},
{
"epoch": 0.8552245616322274,
"grad_norm": 0.35195401310920715,
"learning_rate": 0.00014251993401134146,
"loss": 0.6306336879730224,
"step": 3280
},
{
"epoch": 0.8578319535884232,
"grad_norm": 0.3122062683105469,
"learning_rate": 0.00014104493415784686,
"loss": 0.5940380096435547,
"step": 3290
},
{
"epoch": 0.860439345544619,
"grad_norm": 0.36843088269233704,
"learning_rate": 0.0001395746015767423,
"loss": 0.6059514045715332,
"step": 3300
},
{
"epoch": 0.8630467375008148,
"grad_norm": 0.38153746724128723,
"learning_rate": 0.00013810899925218207,
"loss": 0.6439221858978271,
"step": 3310
},
{
"epoch": 0.8656541294570106,
"grad_norm": 0.34885868430137634,
"learning_rate": 0.0001366481899656916,
"loss": 0.6030439376831055,
"step": 3320
},
{
"epoch": 0.8682615214132064,
"grad_norm": 0.33361926674842834,
"learning_rate": 0.00013519223629347889,
"loss": 0.6526508331298828,
"step": 3330
},
{
"epoch": 0.8708689133694023,
"grad_norm": 0.34130218625068665,
"learning_rate": 0.0001337412006037534,
"loss": 0.6068764686584472,
"step": 3340
},
{
"epoch": 0.8734763053255981,
"grad_norm": 0.29833903908729553,
"learning_rate": 0.00013229514505405478,
"loss": 0.6313079833984375,
"step": 3350
},
{
"epoch": 0.8760836972817939,
"grad_norm": 0.3103373646736145,
"learning_rate": 0.00013085413158858987,
"loss": 0.608611011505127,
"step": 3360
},
{
"epoch": 0.8786910892379897,
"grad_norm": 0.3556700348854065,
"learning_rate": 0.00012941822193557944,
"loss": 0.6170370578765869,
"step": 3370
},
{
"epoch": 0.8812984811941855,
"grad_norm": 0.3218812644481659,
"learning_rate": 0.00012798747760461399,
"loss": 0.6098412036895752,
"step": 3380
},
{
"epoch": 0.8839058731503814,
"grad_norm": 0.3217892348766327,
"learning_rate": 0.00012656195988401876,
"loss": 0.614872932434082,
"step": 3390
},
{
"epoch": 0.8865132651065771,
"grad_norm": 0.35425886511802673,
"learning_rate": 0.0001251417298382284,
"loss": 0.5770713329315186,
"step": 3400
},
{
"epoch": 0.889120657062773,
"grad_norm": 0.36001381278038025,
"learning_rate": 0.00012372684830517117,
"loss": 0.6112618446350098,
"step": 3410
},
{
"epoch": 0.8917280490189687,
"grad_norm": 0.34877336025238037,
"learning_rate": 0.0001223173758936629,
"loss": 0.5867466926574707,
"step": 3420
},
{
"epoch": 0.8943354409751646,
"grad_norm": 0.3069165050983429,
"learning_rate": 0.00012091337298081054,
"loss": 0.5738804340362549,
"step": 3430
},
{
"epoch": 0.8969428329313605,
"grad_norm": 0.3275323510169983,
"learning_rate": 0.00011951489970942614,
"loss": 0.6496532440185547,
"step": 3440
},
{
"epoch": 0.8995502248875562,
"grad_norm": 0.28583991527557373,
"learning_rate": 0.00011812201598544998,
"loss": 0.5960548400878907,
"step": 3450
},
{
"epoch": 0.9021576168437521,
"grad_norm": 0.34295573830604553,
"learning_rate": 0.00011673478147538497,
"loss": 0.5728883266448974,
"step": 3460
},
{
"epoch": 0.9047650087999478,
"grad_norm": 0.3656381368637085,
"learning_rate": 0.00011535325560374049,
"loss": 0.5950040340423584,
"step": 3470
},
{
"epoch": 0.9073724007561437,
"grad_norm": 0.34811246395111084,
"learning_rate": 0.00011397749755048678,
"loss": 0.5900771617889404,
"step": 3480
},
{
"epoch": 0.9099797927123395,
"grad_norm": 0.312678724527359,
"learning_rate": 0.00011260756624851997,
"loss": 0.5731687068939209,
"step": 3490
},
{
"epoch": 0.9125871846685353,
"grad_norm": 0.38274896144866943,
"learning_rate": 0.00011124352038113755,
"loss": 0.5754813194274903,
"step": 3500
},
{
"epoch": 0.9125871846685353,
"eval_loss": 2.32708740234375,
"eval_runtime": 60.7839,
"eval_samples_per_second": 9.739,
"eval_steps_per_second": 1.217,
"step": 3500
},
{
"epoch": 0.9151945766247311,
"grad_norm": 0.26276397705078125,
"learning_rate": 0.00010988541837952457,
"loss": 0.5769128799438477,
"step": 3510
},
{
"epoch": 0.9178019685809269,
"grad_norm": 0.3297964930534363,
"learning_rate": 0.0001085333184202508,
"loss": 0.6111959457397461,
"step": 3520
},
{
"epoch": 0.9204093605371227,
"grad_norm": 0.329017698764801,
"learning_rate": 0.00010718727842277812,
"loss": 0.6030857563018799,
"step": 3530
},
{
"epoch": 0.9230167524933186,
"grad_norm": 0.3764004409313202,
"learning_rate": 0.00010584735604698007,
"loss": 0.6020639896392822,
"step": 3540
},
{
"epoch": 0.9256241444495144,
"grad_norm": 0.39819231629371643,
"learning_rate": 0.00010451360869067183,
"loss": 0.6173338890075684,
"step": 3550
},
{
"epoch": 0.9282315364057102,
"grad_norm": 0.35425636172294617,
"learning_rate": 0.00010318609348715096,
"loss": 0.5878684520721436,
"step": 3560
},
{
"epoch": 0.930838928361906,
"grad_norm": 0.33478814363479614,
"learning_rate": 0.00010186486730275049,
"loss": 0.5536652088165284,
"step": 3570
},
{
"epoch": 0.9334463203181018,
"grad_norm": 0.37404435873031616,
"learning_rate": 0.00010054998673440286,
"loss": 0.6117733478546142,
"step": 3580
},
{
"epoch": 0.9360537122742977,
"grad_norm": 0.3885613977909088,
"learning_rate": 9.924150810721516e-05,
"loss": 0.5528316020965576,
"step": 3590
},
{
"epoch": 0.9386611042304934,
"grad_norm": 0.44533368945121765,
"learning_rate": 9.793948747205694e-05,
"loss": 0.5793837070465088,
"step": 3600
},
{
"epoch": 0.9412684961866893,
"grad_norm": 0.41212591528892517,
"learning_rate": 9.664398060315876e-05,
"loss": 0.5612047672271728,
"step": 3610
},
{
"epoch": 0.943875888142885,
"grad_norm": 0.3158949315547943,
"learning_rate": 9.535504299572307e-05,
"loss": 0.570124340057373,
"step": 3620
},
{
"epoch": 0.9464832800990809,
"grad_norm": 0.3416590988636017,
"learning_rate": 9.407272986354703e-05,
"loss": 0.5875216484069824,
"step": 3630
},
{
"epoch": 0.9490906720552768,
"grad_norm": 0.3240266740322113,
"learning_rate": 9.279709613665743e-05,
"loss": 0.5980135440826416,
"step": 3640
},
{
"epoch": 0.9516980640114725,
"grad_norm": 0.29565727710723877,
"learning_rate": 9.152819645895752e-05,
"loss": 0.6013434410095215,
"step": 3650
},
{
"epoch": 0.9543054559676684,
"grad_norm": 0.32616135478019714,
"learning_rate": 9.026608518588612e-05,
"loss": 0.5690362930297852,
"step": 3660
},
{
"epoch": 0.9569128479238641,
"grad_norm": 0.3920518755912781,
"learning_rate": 8.90108163820896e-05,
"loss": 0.5698985099792481,
"step": 3670
},
{
"epoch": 0.95952023988006,
"grad_norm": 0.2985530197620392,
"learning_rate": 8.776244381910567e-05,
"loss": 0.5808091640472413,
"step": 3680
},
{
"epoch": 0.9621276318362558,
"grad_norm": 0.31916192173957825,
"learning_rate": 8.652102097305994e-05,
"loss": 0.5854286670684814,
"step": 3690
},
{
"epoch": 0.9647350237924516,
"grad_norm": 0.33721548318862915,
"learning_rate": 8.528660102237551e-05,
"loss": 0.5722479820251465,
"step": 3700
},
{
"epoch": 0.9673424157486474,
"grad_norm": 0.38025516271591187,
"learning_rate": 8.405923684549464e-05,
"loss": 0.5725702762603759,
"step": 3710
},
{
"epoch": 0.9699498077048432,
"grad_norm": 0.3271743655204773,
"learning_rate": 8.283898101861345e-05,
"loss": 0.5750481128692627,
"step": 3720
},
{
"epoch": 0.972557199661039,
"grad_norm": 0.38187846541404724,
"learning_rate": 8.162588581343036e-05,
"loss": 0.5244960308074951,
"step": 3730
},
{
"epoch": 0.9751645916172349,
"grad_norm": 0.34241780638694763,
"learning_rate": 8.04200031949065e-05,
"loss": 0.542147159576416,
"step": 3740
},
{
"epoch": 0.9777719835734306,
"grad_norm": 0.33918023109436035,
"learning_rate": 7.922138481903968e-05,
"loss": 0.5492253303527832,
"step": 3750
},
{
"epoch": 0.9803793755296265,
"grad_norm": 0.32857686281204224,
"learning_rate": 7.803008203065187e-05,
"loss": 0.5587058067321777,
"step": 3760
},
{
"epoch": 0.9829867674858223,
"grad_norm": 0.34119388461112976,
"learning_rate": 7.684614586118954e-05,
"loss": 0.5815017223358154,
"step": 3770
},
{
"epoch": 0.9855941594420181,
"grad_norm": 0.33758214116096497,
"learning_rate": 7.566962702653781e-05,
"loss": 0.5678920745849609,
"step": 3780
},
{
"epoch": 0.988201551398214,
"grad_norm": 0.3500860631465912,
"learning_rate": 7.450057592484768e-05,
"loss": 0.5604970455169678,
"step": 3790
},
{
"epoch": 0.9908089433544097,
"grad_norm": 0.3810959756374359,
"learning_rate": 7.333904263437748e-05,
"loss": 0.5577153205871582,
"step": 3800
},
{
"epoch": 0.9934163353106056,
"grad_norm": 0.3265530467033386,
"learning_rate": 7.218507691134752e-05,
"loss": 0.5658472061157227,
"step": 3810
},
{
"epoch": 0.9960237272668013,
"grad_norm": 0.3351763188838959,
"learning_rate": 7.103872818780866e-05,
"loss": 0.5463396072387695,
"step": 3820
},
{
"epoch": 0.9986311192229972,
"grad_norm": 0.3470080494880676,
"learning_rate": 6.990004556952487e-05,
"loss": 0.5723049640655518,
"step": 3830
},
{
"epoch": 1.0010429567824783,
"grad_norm": 0.3287046253681183,
"learning_rate": 6.876907783386971e-05,
"loss": 0.5544593334197998,
"step": 3840
},
{
"epoch": 1.003650348738674,
"grad_norm": 0.3052399456501007,
"learning_rate": 6.764587342773679e-05,
"loss": 0.49776530265808105,
"step": 3850
},
{
"epoch": 1.00625774069487,
"grad_norm": 0.348078191280365,
"learning_rate": 6.653048046546456e-05,
"loss": 0.47914881706237794,
"step": 3860
},
{
"epoch": 1.0088651326510658,
"grad_norm": 0.36548739671707153,
"learning_rate": 6.542294672677526e-05,
"loss": 0.4768219947814941,
"step": 3870
},
{
"epoch": 1.0114725246072616,
"grad_norm": 0.31463202834129333,
"learning_rate": 6.432331965472801e-05,
"loss": 0.48053803443908694,
"step": 3880
},
{
"epoch": 1.0140799165634573,
"grad_norm": 0.3365277647972107,
"learning_rate": 6.32316463536868e-05,
"loss": 0.469533109664917,
"step": 3890
},
{
"epoch": 1.0166873085196533,
"grad_norm": 0.3088042140007019,
"learning_rate": 6.214797358730248e-05,
"loss": 0.45445098876953127,
"step": 3900
},
{
"epoch": 1.019294700475849,
"grad_norm": 0.37573444843292236,
"learning_rate": 6.107234777650963e-05,
"loss": 0.49057722091674805,
"step": 3910
},
{
"epoch": 1.0219020924320448,
"grad_norm": 0.3293447494506836,
"learning_rate": 6.000481499753793e-05,
"loss": 0.48319287300109864,
"step": 3920
},
{
"epoch": 1.0245094843882407,
"grad_norm": 0.2775663733482361,
"learning_rate": 5.8945420979938686e-05,
"loss": 0.4844508171081543,
"step": 3930
},
{
"epoch": 1.0271168763444365,
"grad_norm": 0.3209178149700165,
"learning_rate": 5.7894211104625614e-05,
"loss": 0.4774619102478027,
"step": 3940
},
{
"epoch": 1.0297242683006322,
"grad_norm": 0.30637258291244507,
"learning_rate": 5.685123040193113e-05,
"loss": 0.45171213150024414,
"step": 3950
},
{
"epoch": 1.0323316602568282,
"grad_norm": 0.3400314152240753,
"learning_rate": 5.5816523549677226e-05,
"loss": 0.4796245574951172,
"step": 3960
},
{
"epoch": 1.034939052213024,
"grad_norm": 0.3137646019458771,
"learning_rate": 5.4790134871261636e-05,
"loss": 0.4850926399230957,
"step": 3970
},
{
"epoch": 1.0375464441692197,
"grad_norm": 0.345639169216156,
"learning_rate": 5.3772108333759314e-05,
"loss": 0.45691981315612795,
"step": 3980
},
{
"epoch": 1.0401538361254155,
"grad_norm": 0.44569915533065796,
"learning_rate": 5.2762487546038854e-05,
"loss": 0.4801798820495605,
"step": 3990
},
{
"epoch": 1.0427612280816114,
"grad_norm": 0.33535394072532654,
"learning_rate": 5.176131575689447e-05,
"loss": 0.5038554191589355,
"step": 4000
},
{
"epoch": 1.0427612280816114,
"eval_loss": 2.4464287757873535,
"eval_runtime": 60.9424,
"eval_samples_per_second": 9.714,
"eval_steps_per_second": 1.214,
"step": 4000
},
{
"epoch": 1.0453686200378072,
"grad_norm": 0.3068539500236511,
"learning_rate": 5.076863585319352e-05,
"loss": 0.45174031257629393,
"step": 4010
},
{
"epoch": 1.047976011994003,
"grad_norm": 0.3753582537174225,
"learning_rate": 4.978449035803917e-05,
"loss": 0.4762752056121826,
"step": 4020
},
{
"epoch": 1.050583403950199,
"grad_norm": 0.36191460490226746,
"learning_rate": 4.880892142894891e-05,
"loss": 0.48446078300476075,
"step": 4030
},
{
"epoch": 1.0531907959063946,
"grad_norm": 0.3330017626285553,
"learning_rate": 4.78419708560488e-05,
"loss": 0.46834425926208495,
"step": 4040
},
{
"epoch": 1.0557981878625904,
"grad_norm": 0.32094594836235046,
"learning_rate": 4.688368006028299e-05,
"loss": 0.4974546432495117,
"step": 4050
},
{
"epoch": 1.0584055798187864,
"grad_norm": 0.3017931580543518,
"learning_rate": 4.593409009163982e-05,
"loss": 0.46979737281799316,
"step": 4060
},
{
"epoch": 1.061012971774982,
"grad_norm": 0.36359062790870667,
"learning_rate": 4.4993241627393015e-05,
"loss": 0.49051923751831056,
"step": 4070
},
{
"epoch": 1.0636203637311779,
"grad_norm": 0.3066091239452362,
"learning_rate": 4.406117497035944e-05,
"loss": 0.46471424102783204,
"step": 4080
},
{
"epoch": 1.0662277556873736,
"grad_norm": 0.3397349417209625,
"learning_rate": 4.3137930047172506e-05,
"loss": 0.43993582725524905,
"step": 4090
},
{
"epoch": 1.0688351476435696,
"grad_norm": 0.36894890666007996,
"learning_rate": 4.2223546406571944e-05,
"loss": 0.4743666172027588,
"step": 4100
},
{
"epoch": 1.0714425395997653,
"grad_norm": 0.3282569646835327,
"learning_rate": 4.131806321770959e-05,
"loss": 0.473482608795166,
"step": 4110
},
{
"epoch": 1.074049931555961,
"grad_norm": 0.28172239661216736,
"learning_rate": 4.0421519268471624e-05,
"loss": 0.43282437324523926,
"step": 4120
},
{
"epoch": 1.076657323512157,
"grad_norm": 0.33178994059562683,
"learning_rate": 3.9533952963816815e-05,
"loss": 0.4629314422607422,
"step": 4130
},
{
"epoch": 1.0792647154683528,
"grad_norm": 0.3200715482234955,
"learning_rate": 3.865540232413156e-05,
"loss": 0.4904191970825195,
"step": 4140
},
{
"epoch": 1.0818721074245485,
"grad_norm": 0.3135477900505066,
"learning_rate": 3.77859049836014e-05,
"loss": 0.4488736629486084,
"step": 4150
},
{
"epoch": 1.0844794993807445,
"grad_norm": 0.3498687744140625,
"learning_rate": 3.692549818859839e-05,
"loss": 0.4469603538513184,
"step": 4160
},
{
"epoch": 1.0870868913369403,
"grad_norm": 0.31335851550102234,
"learning_rate": 3.6074218796085966e-05,
"loss": 0.4704417705535889,
"step": 4170
},
{
"epoch": 1.089694283293136,
"grad_norm": 0.34564676880836487,
"learning_rate": 3.52321032720401e-05,
"loss": 0.4614598751068115,
"step": 4180
},
{
"epoch": 1.092301675249332,
"grad_norm": 0.3244510889053345,
"learning_rate": 3.439918768988695e-05,
"loss": 0.47153382301330565,
"step": 4190
},
{
"epoch": 1.0949090672055277,
"grad_norm": 0.271252304315567,
"learning_rate": 3.3575507728957926e-05,
"loss": 0.4237504959106445,
"step": 4200
},
{
"epoch": 1.0975164591617235,
"grad_norm": 0.3548557162284851,
"learning_rate": 3.276109867296117e-05,
"loss": 0.46147937774658204,
"step": 4210
},
{
"epoch": 1.1001238511179192,
"grad_norm": 0.36666133999824524,
"learning_rate": 3.195599540847008e-05,
"loss": 0.47422361373901367,
"step": 4220
},
{
"epoch": 1.1027312430741152,
"grad_norm": 0.32674500346183777,
"learning_rate": 3.1160232423428885e-05,
"loss": 0.46683483123779296,
"step": 4230
},
{
"epoch": 1.105338635030311,
"grad_norm": 0.3393232524394989,
"learning_rate": 3.03738438056754e-05,
"loss": 0.46091880798339846,
"step": 4240
},
{
"epoch": 1.1079460269865067,
"grad_norm": 0.35054823756217957,
"learning_rate": 2.9596863241480697e-05,
"loss": 0.4754704475402832,
"step": 4250
},
{
"epoch": 1.1105534189427027,
"grad_norm": 0.30949196219444275,
"learning_rate": 2.8829324014106122e-05,
"loss": 0.4519375801086426,
"step": 4260
},
{
"epoch": 1.1131608108988984,
"grad_norm": 0.3263954520225525,
"learning_rate": 2.8071259002377585e-05,
"loss": 0.4598262786865234,
"step": 4270
},
{
"epoch": 1.1157682028550941,
"grad_norm": 0.34200191497802734,
"learning_rate": 2.7322700679277223e-05,
"loss": 0.4622962951660156,
"step": 4280
},
{
"epoch": 1.11837559481129,
"grad_norm": 0.33595728874206543,
"learning_rate": 2.6583681110552093e-05,
"loss": 0.45746655464172364,
"step": 4290
},
{
"epoch": 1.1209829867674859,
"grad_norm": 0.37495556473731995,
"learning_rate": 2.5854231953340995e-05,
"loss": 0.4772180557250977,
"step": 4300
},
{
"epoch": 1.1235903787236816,
"grad_norm": 0.3797667324542999,
"learning_rate": 2.5134384454818028e-05,
"loss": 0.47903079986572267,
"step": 4310
},
{
"epoch": 1.1261977706798774,
"grad_norm": 0.2765319347381592,
"learning_rate": 2.44241694508541e-05,
"loss": 0.4512158393859863,
"step": 4320
},
{
"epoch": 1.1288051626360733,
"grad_norm": 0.35347455739974976,
"learning_rate": 2.3723617364696247e-05,
"loss": 0.46738595962524415,
"step": 4330
},
{
"epoch": 1.131412554592269,
"grad_norm": 0.30416563153266907,
"learning_rate": 2.3032758205664224e-05,
"loss": 0.4623900890350342,
"step": 4340
},
{
"epoch": 1.1340199465484648,
"grad_norm": 0.3307722806930542,
"learning_rate": 2.2351621567865022e-05,
"loss": 0.4654539585113525,
"step": 4350
},
{
"epoch": 1.1366273385046608,
"grad_norm": 0.4208446741104126,
"learning_rate": 2.1680236628925238e-05,
"loss": 0.4713085174560547,
"step": 4360
},
{
"epoch": 1.1392347304608565,
"grad_norm": 0.30065780878067017,
"learning_rate": 2.1018632148741113e-05,
"loss": 0.481781530380249,
"step": 4370
},
{
"epoch": 1.1418421224170523,
"grad_norm": 0.3753151297569275,
"learning_rate": 2.036683646824658e-05,
"loss": 0.4670425891876221,
"step": 4380
},
{
"epoch": 1.1444495143732483,
"grad_norm": 0.3752826452255249,
"learning_rate": 1.9724877508199235e-05,
"loss": 0.48926310539245604,
"step": 4390
},
{
"epoch": 1.147056906329444,
"grad_norm": 0.33213159441947937,
"learning_rate": 1.909278276798429e-05,
"loss": 0.45664496421813966,
"step": 4400
},
{
"epoch": 1.1496642982856398,
"grad_norm": 0.3351505398750305,
"learning_rate": 1.8470579324436714e-05,
"loss": 0.46279339790344237,
"step": 4410
},
{
"epoch": 1.1522716902418355,
"grad_norm": 0.3027290403842926,
"learning_rate": 1.7858293830681104e-05,
"loss": 0.465179443359375,
"step": 4420
},
{
"epoch": 1.1548790821980315,
"grad_norm": 0.3453589379787445,
"learning_rate": 1.7255952514990236e-05,
"loss": 0.4608203411102295,
"step": 4430
},
{
"epoch": 1.1574864741542272,
"grad_norm": 0.31773585081100464,
"learning_rate": 1.666358117966127e-05,
"loss": 0.48364691734313964,
"step": 4440
},
{
"epoch": 1.160093866110423,
"grad_norm": 0.3335360586643219,
"learning_rate": 1.6081205199910687e-05,
"loss": 0.4583646297454834,
"step": 4450
},
{
"epoch": 1.162701258066619,
"grad_norm": 0.33595699071884155,
"learning_rate": 1.550884952278714e-05,
"loss": 0.4706264495849609,
"step": 4460
},
{
"epoch": 1.1653086500228147,
"grad_norm": 0.338228315114975,
"learning_rate": 1.4946538666102877e-05,
"loss": 0.4724998474121094,
"step": 4470
},
{
"epoch": 1.1679160419790104,
"grad_norm": 0.3023951053619385,
"learning_rate": 1.4394296717383516e-05,
"loss": 0.4300068378448486,
"step": 4480
},
{
"epoch": 1.1705234339352062,
"grad_norm": 0.36846086382865906,
"learning_rate": 1.385214733283613e-05,
"loss": 0.47844820022583007,
"step": 4490
},
{
"epoch": 1.1731308258914022,
"grad_norm": 0.32578301429748535,
"learning_rate": 1.3320113736335931e-05,
"loss": 0.47261805534362794,
"step": 4500
},
{
"epoch": 1.1731308258914022,
"eval_loss": 2.505293607711792,
"eval_runtime": 60.7436,
"eval_samples_per_second": 9.746,
"eval_steps_per_second": 1.218,
"step": 4500
},
{
"epoch": 1.175738217847598,
"grad_norm": 0.33165180683135986,
"learning_rate": 1.2798218718431453e-05,
"loss": 0.45057263374328616,
"step": 4510
},
{
"epoch": 1.1783456098037937,
"grad_norm": 0.3142957389354706,
"learning_rate": 1.2286484635368244e-05,
"loss": 0.47219123840332033,
"step": 4520
},
{
"epoch": 1.1809530017599896,
"grad_norm": 0.3821601867675781,
"learning_rate": 1.1784933408131248e-05,
"loss": 0.4850562572479248,
"step": 4530
},
{
"epoch": 1.1835603937161854,
"grad_norm": 0.34352943301200867,
"learning_rate": 1.129358652150575e-05,
"loss": 0.46318879127502444,
"step": 4540
},
{
"epoch": 1.1861677856723811,
"grad_norm": 0.29660624265670776,
"learning_rate": 1.0812465023157025e-05,
"loss": 0.46358489990234375,
"step": 4550
},
{
"epoch": 1.188775177628577,
"grad_norm": 0.34422022104263306,
"learning_rate": 1.034158952272879e-05,
"loss": 0.44614782333374026,
"step": 4560
},
{
"epoch": 1.1913825695847728,
"grad_norm": 0.3352455496788025,
"learning_rate": 9.880980190960315e-06,
"loss": 0.45110330581665037,
"step": 4570
},
{
"epoch": 1.1939899615409686,
"grad_norm": 0.25852876901626587,
"learning_rate": 9.430656758822264e-06,
"loss": 0.4344322681427002,
"step": 4580
},
{
"epoch": 1.1965973534971646,
"grad_norm": 0.36658501625061035,
"learning_rate": 8.990638516671757e-06,
"loss": 0.45212082862854003,
"step": 4590
},
{
"epoch": 1.1992047454533603,
"grad_norm": 0.32412490248680115,
"learning_rate": 8.560944313425772e-06,
"loss": 0.4529105663299561,
"step": 4600
},
{
"epoch": 1.201812137409556,
"grad_norm": 0.2986154556274414,
"learning_rate": 8.141592555753807e-06,
"loss": 0.4528106689453125,
"step": 4610
},
{
"epoch": 1.2044195293657518,
"grad_norm": 0.322527140378952,
"learning_rate": 7.732601207289497e-06,
"loss": 0.4522389888763428,
"step": 4620
},
{
"epoch": 1.2070269213219478,
"grad_norm": 0.2830180823802948,
"learning_rate": 7.333987787860958e-06,
"loss": 0.44110922813415526,
"step": 4630
},
{
"epoch": 1.2096343132781435,
"grad_norm": 0.3009679317474365,
"learning_rate": 6.945769372740413e-06,
"loss": 0.43086953163146974,
"step": 4640
},
{
"epoch": 1.2122417052343393,
"grad_norm": 0.31641507148742676,
"learning_rate": 6.567962591912613e-06,
"loss": 0.45824193954467773,
"step": 4650
},
{
"epoch": 1.2148490971905352,
"grad_norm": 0.3088202178478241,
"learning_rate": 6.200583629362683e-06,
"loss": 0.4743471145629883,
"step": 4660
},
{
"epoch": 1.217456489146731,
"grad_norm": 0.31374308466911316,
"learning_rate": 5.843648222382591e-06,
"loss": 0.43962993621826174,
"step": 4670
},
{
"epoch": 1.2200638811029267,
"grad_norm": 0.3322969377040863,
"learning_rate": 5.497171660897221e-06,
"loss": 0.4515493869781494,
"step": 4680
},
{
"epoch": 1.2226712730591227,
"grad_norm": 0.34848204255104065,
"learning_rate": 5.161168786809334e-06,
"loss": 0.4837063789367676,
"step": 4690
},
{
"epoch": 1.2252786650153185,
"grad_norm": 0.35345688462257385,
"learning_rate": 4.835653993363693e-06,
"loss": 0.44669346809387206,
"step": 4700
},
{
"epoch": 1.2278860569715142,
"grad_norm": 0.33362483978271484,
"learning_rate": 4.5206412245306916e-06,
"loss": 0.44629697799682616,
"step": 4710
},
{
"epoch": 1.2304934489277102,
"grad_norm": 0.3215402066707611,
"learning_rate": 4.216143974408892e-06,
"loss": 0.4523158073425293,
"step": 4720
},
{
"epoch": 1.233100840883906,
"grad_norm": 0.28513166308403015,
"learning_rate": 3.922175286647067e-06,
"loss": 0.4632432460784912,
"step": 4730
},
{
"epoch": 1.2357082328401017,
"grad_norm": 0.2997485101222992,
"learning_rate": 3.6387477538853132e-06,
"loss": 0.43720459938049316,
"step": 4740
},
{
"epoch": 1.2383156247962974,
"grad_norm": 0.32336780428886414,
"learning_rate": 3.3658735172159284e-06,
"loss": 0.45301198959350586,
"step": 4750
},
{
"epoch": 1.2409230167524934,
"grad_norm": 0.3227427899837494,
"learning_rate": 3.1035642656629393e-06,
"loss": 0.4496479511260986,
"step": 4760
},
{
"epoch": 1.2435304087086891,
"grad_norm": 0.276816189289093,
"learning_rate": 2.851831235681668e-06,
"loss": 0.4664166927337646,
"step": 4770
},
{
"epoch": 1.2461378006648849,
"grad_norm": 0.3105098009109497,
"learning_rate": 2.6106852106773106e-06,
"loss": 0.4325725555419922,
"step": 4780
},
{
"epoch": 1.2487451926210809,
"grad_norm": 0.3525710701942444,
"learning_rate": 2.380136520542947e-06,
"loss": 0.47321524620056155,
"step": 4790
},
{
"epoch": 1.2513525845772766,
"grad_norm": 0.32424336671829224,
"learning_rate": 2.1601950412171723e-06,
"loss": 0.46515703201293945,
"step": 4800
},
{
"epoch": 1.2539599765334724,
"grad_norm": 0.29163214564323425,
"learning_rate": 1.9508701942609064e-06,
"loss": 0.46263608932495115,
"step": 4810
},
{
"epoch": 1.256567368489668,
"grad_norm": 0.3159816265106201,
"learning_rate": 1.7521709464539127e-06,
"loss": 0.46936697959899903,
"step": 4820
},
{
"epoch": 1.259174760445864,
"grad_norm": 0.3383216857910156,
"learning_rate": 1.564105809410632e-06,
"loss": 0.45978994369506837,
"step": 4830
},
{
"epoch": 1.2617821524020598,
"grad_norm": 0.3204449713230133,
"learning_rate": 1.3866828392156416e-06,
"loss": 0.4596550464630127,
"step": 4840
},
{
"epoch": 1.2643895443582558,
"grad_norm": 0.34576305747032166,
"learning_rate": 1.219909636078459e-06,
"loss": 0.46024765968322756,
"step": 4850
},
{
"epoch": 1.2669969363144515,
"grad_norm": 0.31107112765312195,
"learning_rate": 1.0637933440080516e-06,
"loss": 0.4407492637634277,
"step": 4860
},
{
"epoch": 1.2696043282706473,
"grad_norm": 0.322909951210022,
"learning_rate": 9.183406505067494e-07,
"loss": 0.4623416423797607,
"step": 4870
},
{
"epoch": 1.272211720226843,
"grad_norm": 0.3140254318714142,
"learning_rate": 7.835577862838339e-07,
"loss": 0.4614372730255127,
"step": 4880
},
{
"epoch": 1.2748191121830388,
"grad_norm": 0.32432618737220764,
"learning_rate": 6.594505249886695e-07,
"loss": 0.4283143520355225,
"step": 4890
},
{
"epoch": 1.2774265041392348,
"grad_norm": 0.3526235818862915,
"learning_rate": 5.460241829632063e-07,
"loss": 0.47379159927368164,
"step": 4900
},
{
"epoch": 1.2800338960954305,
"grad_norm": 0.31688305735588074,
"learning_rate": 4.432836190144951e-07,
"loss": 0.42535972595214844,
"step": 4910
},
{
"epoch": 1.2826412880516265,
"grad_norm": 0.31910261511802673,
"learning_rate": 3.5123323420627183e-07,
"loss": 0.4553154468536377,
"step": 4920
},
{
"epoch": 1.2852486800078222,
"grad_norm": 0.30769890546798706,
"learning_rate": 2.698769716706906e-07,
"loss": 0.45520939826965334,
"step": 4930
},
{
"epoch": 1.287856071964018,
"grad_norm": 0.3247165381908417,
"learning_rate": 1.9921831643926468e-07,
"loss": 0.453525972366333,
"step": 4940
},
{
"epoch": 1.2904634639202137,
"grad_norm": 0.3217741847038269,
"learning_rate": 1.3926029529362506e-07,
"loss": 0.45581507682800293,
"step": 4950
},
{
"epoch": 1.2930708558764097,
"grad_norm": 0.3457319140434265,
"learning_rate": 9.000547663581826e-08,
"loss": 0.46144566535949705,
"step": 4960
},
{
"epoch": 1.2956782478326054,
"grad_norm": 0.2856600880622864,
"learning_rate": 5.145597037833904e-08,
"loss": 0.4681571960449219,
"step": 4970
},
{
"epoch": 1.2982856397888012,
"grad_norm": 0.27411606907844543,
"learning_rate": 2.3613427853758086e-08,
"loss": 0.4496706485748291,
"step": 4980
},
{
"epoch": 1.3008930317449972,
"grad_norm": 0.3205054998397827,
"learning_rate": 6.479041743945313e-09,
"loss": 0.45045809745788573,
"step": 4990
},
{
"epoch": 1.303500423701193,
"grad_norm": 0.28665968775749207,
"learning_rate": 5.354602891638471e-11,
"loss": 0.44310483932495115,
"step": 5000
},
{
"epoch": 1.303500423701193,
"eval_loss": 2.518051862716675,
"eval_runtime": 60.7004,
"eval_samples_per_second": 9.753,
"eval_steps_per_second": 1.219,
"step": 5000
}
],
"logging_steps": 10,
"max_steps": 5000,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 1000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 1.0438201745119642e+17,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}