Safetensors
MMPT-FM / MMPT_FM /trainer_state.json
model-ingest's picture
Upload 36 files
43cdbee verified
{
"best_metric": 0.3424564003944397,
"best_model_checkpoint": "./results/chembl.trans_only_T5Chem_dmodel512_layers12_batch120_lr0.0005_train0.80/checkpoint-25000",
"epoch": 1.4214236979758925,
"eval_steps": 5000,
"global_step": 25000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.002842847395951785,
"grad_norm": 0.4646081328392029,
"learning_rate": 0.0004999955580509438,
"loss": 0.6585,
"step": 50
},
{
"epoch": 0.00568569479190357,
"grad_norm": 0.19792014360427856,
"learning_rate": 0.0004999911161018877,
"loss": 0.53,
"step": 100
},
{
"epoch": 0.008528542187855356,
"grad_norm": 0.16343972086906433,
"learning_rate": 0.0004999866741528315,
"loss": 0.5034,
"step": 150
},
{
"epoch": 0.01137138958380714,
"grad_norm": 0.20607765018939972,
"learning_rate": 0.0004999822322037754,
"loss": 0.496,
"step": 200
},
{
"epoch": 0.014214236979758927,
"grad_norm": 0.14076846837997437,
"learning_rate": 0.0004999777902547192,
"loss": 0.4841,
"step": 250
},
{
"epoch": 0.017057084375710713,
"grad_norm": 0.12710894644260406,
"learning_rate": 0.000499973348305663,
"loss": 0.4816,
"step": 300
},
{
"epoch": 0.019899931771662497,
"grad_norm": 0.1586833894252777,
"learning_rate": 0.0004999689063566068,
"loss": 0.477,
"step": 350
},
{
"epoch": 0.02274277916761428,
"grad_norm": 0.18906599283218384,
"learning_rate": 0.0004999644644075506,
"loss": 0.4743,
"step": 400
},
{
"epoch": 0.02558562656356607,
"grad_norm": 0.14747406542301178,
"learning_rate": 0.0004999600224584944,
"loss": 0.465,
"step": 450
},
{
"epoch": 0.028428473959517853,
"grad_norm": 0.15983326733112335,
"learning_rate": 0.0004999555805094383,
"loss": 0.4647,
"step": 500
},
{
"epoch": 0.03127132135546964,
"grad_norm": 0.14171165227890015,
"learning_rate": 0.0004999511385603821,
"loss": 0.4603,
"step": 550
},
{
"epoch": 0.034114168751421425,
"grad_norm": 0.13967591524124146,
"learning_rate": 0.0004999466966113259,
"loss": 0.4574,
"step": 600
},
{
"epoch": 0.036957016147373206,
"grad_norm": 0.2039215862751007,
"learning_rate": 0.0004999422546622698,
"loss": 0.4536,
"step": 650
},
{
"epoch": 0.039799863543324994,
"grad_norm": 0.15800215303897858,
"learning_rate": 0.0004999378127132136,
"loss": 0.4544,
"step": 700
},
{
"epoch": 0.04264271093927678,
"grad_norm": 0.16743983328342438,
"learning_rate": 0.0004999333707641574,
"loss": 0.4528,
"step": 750
},
{
"epoch": 0.04548555833522856,
"grad_norm": 0.16410502791404724,
"learning_rate": 0.0004999289288151012,
"loss": 0.4498,
"step": 800
},
{
"epoch": 0.04832840573118035,
"grad_norm": 0.17722439765930176,
"learning_rate": 0.000499924486866045,
"loss": 0.4461,
"step": 850
},
{
"epoch": 0.05117125312713214,
"grad_norm": 0.13122214376926422,
"learning_rate": 0.0004999200449169889,
"loss": 0.4422,
"step": 900
},
{
"epoch": 0.05401410052308392,
"grad_norm": 0.1328468769788742,
"learning_rate": 0.0004999156029679327,
"loss": 0.4438,
"step": 950
},
{
"epoch": 0.05685694791903571,
"grad_norm": 0.1839675009250641,
"learning_rate": 0.0004999111610188765,
"loss": 0.4429,
"step": 1000
},
{
"epoch": 0.059699795314987494,
"grad_norm": 0.14567476511001587,
"learning_rate": 0.0004999067190698203,
"loss": 0.4356,
"step": 1050
},
{
"epoch": 0.06254264271093928,
"grad_norm": 0.1527237445116043,
"learning_rate": 0.0004999022771207641,
"loss": 0.438,
"step": 1100
},
{
"epoch": 0.06538549010689106,
"grad_norm": 0.1287076622247696,
"learning_rate": 0.000499897835171708,
"loss": 0.4343,
"step": 1150
},
{
"epoch": 0.06822833750284285,
"grad_norm": 0.11131107807159424,
"learning_rate": 0.0004998933932226518,
"loss": 0.4367,
"step": 1200
},
{
"epoch": 0.07107118489879463,
"grad_norm": 0.13373662531375885,
"learning_rate": 0.0004998889512735956,
"loss": 0.4316,
"step": 1250
},
{
"epoch": 0.07391403229474641,
"grad_norm": 0.13251946866512299,
"learning_rate": 0.0004998845093245395,
"loss": 0.4347,
"step": 1300
},
{
"epoch": 0.07675687969069821,
"grad_norm": 0.14924846589565277,
"learning_rate": 0.0004998800673754833,
"loss": 0.4315,
"step": 1350
},
{
"epoch": 0.07959972708664999,
"grad_norm": 0.12222905457019806,
"learning_rate": 0.0004998756254264271,
"loss": 0.4311,
"step": 1400
},
{
"epoch": 0.08244257448260177,
"grad_norm": 0.10474367439746857,
"learning_rate": 0.0004998711834773709,
"loss": 0.4289,
"step": 1450
},
{
"epoch": 0.08528542187855356,
"grad_norm": 0.11413396149873734,
"learning_rate": 0.0004998667415283148,
"loss": 0.4287,
"step": 1500
},
{
"epoch": 0.08812826927450534,
"grad_norm": 0.1532815843820572,
"learning_rate": 0.0004998622995792587,
"loss": 0.4266,
"step": 1550
},
{
"epoch": 0.09097111667045713,
"grad_norm": 0.16395249962806702,
"learning_rate": 0.0004998578576302025,
"loss": 0.4254,
"step": 1600
},
{
"epoch": 0.09381396406640892,
"grad_norm": 0.13058966398239136,
"learning_rate": 0.0004998534156811463,
"loss": 0.4236,
"step": 1650
},
{
"epoch": 0.0966568114623607,
"grad_norm": 0.15570798516273499,
"learning_rate": 0.0004998489737320901,
"loss": 0.4267,
"step": 1700
},
{
"epoch": 0.09949965885831248,
"grad_norm": 0.1350376456975937,
"learning_rate": 0.0004998445317830339,
"loss": 0.421,
"step": 1750
},
{
"epoch": 0.10234250625426428,
"grad_norm": 0.13523083925247192,
"learning_rate": 0.0004998400898339777,
"loss": 0.4212,
"step": 1800
},
{
"epoch": 0.10518535365021606,
"grad_norm": 0.12544061243534088,
"learning_rate": 0.0004998356478849215,
"loss": 0.424,
"step": 1850
},
{
"epoch": 0.10802820104616784,
"grad_norm": 0.12443090230226517,
"learning_rate": 0.0004998312059358654,
"loss": 0.4217,
"step": 1900
},
{
"epoch": 0.11087104844211963,
"grad_norm": 0.11098171025514603,
"learning_rate": 0.0004998267639868092,
"loss": 0.4226,
"step": 1950
},
{
"epoch": 0.11371389583807141,
"grad_norm": 0.129473477602005,
"learning_rate": 0.000499822322037753,
"loss": 0.4237,
"step": 2000
},
{
"epoch": 0.1165567432340232,
"grad_norm": 0.1107649877667427,
"learning_rate": 0.0004998178800886969,
"loss": 0.4241,
"step": 2050
},
{
"epoch": 0.11939959062997499,
"grad_norm": 0.12195456773042679,
"learning_rate": 0.0004998134381396407,
"loss": 0.4176,
"step": 2100
},
{
"epoch": 0.12224243802592677,
"grad_norm": 0.12097357213497162,
"learning_rate": 0.0004998089961905845,
"loss": 0.4156,
"step": 2150
},
{
"epoch": 0.12508528542187855,
"grad_norm": 0.11261444538831711,
"learning_rate": 0.0004998045542415283,
"loss": 0.4179,
"step": 2200
},
{
"epoch": 0.12792813281783033,
"grad_norm": 0.13488833606243134,
"learning_rate": 0.0004998001122924721,
"loss": 0.4175,
"step": 2250
},
{
"epoch": 0.1307709802137821,
"grad_norm": 0.11204372346401215,
"learning_rate": 0.000499795670343416,
"loss": 0.4169,
"step": 2300
},
{
"epoch": 0.13361382760973392,
"grad_norm": 0.10848614573478699,
"learning_rate": 0.0004997912283943598,
"loss": 0.4156,
"step": 2350
},
{
"epoch": 0.1364566750056857,
"grad_norm": 0.1451840102672577,
"learning_rate": 0.0004997867864453036,
"loss": 0.4154,
"step": 2400
},
{
"epoch": 0.13929952240163748,
"grad_norm": 0.14736251533031464,
"learning_rate": 0.0004997823444962474,
"loss": 0.4155,
"step": 2450
},
{
"epoch": 0.14214236979758926,
"grad_norm": 0.12735013663768768,
"learning_rate": 0.0004997779025471912,
"loss": 0.4089,
"step": 2500
},
{
"epoch": 0.14498521719354104,
"grad_norm": 0.212030827999115,
"learning_rate": 0.0004997734605981351,
"loss": 0.415,
"step": 2550
},
{
"epoch": 0.14782806458949282,
"grad_norm": 0.11738018691539764,
"learning_rate": 0.0004997690186490789,
"loss": 0.4102,
"step": 2600
},
{
"epoch": 0.15067091198544463,
"grad_norm": 0.1060812696814537,
"learning_rate": 0.0004997645767000228,
"loss": 0.4136,
"step": 2650
},
{
"epoch": 0.15351375938139641,
"grad_norm": 0.12909935414791107,
"learning_rate": 0.0004997601347509666,
"loss": 0.4102,
"step": 2700
},
{
"epoch": 0.1563566067773482,
"grad_norm": 0.15052352845668793,
"learning_rate": 0.0004997556928019104,
"loss": 0.4092,
"step": 2750
},
{
"epoch": 0.15919945417329998,
"grad_norm": 0.13438868522644043,
"learning_rate": 0.0004997512508528542,
"loss": 0.4102,
"step": 2800
},
{
"epoch": 0.16204230156925176,
"grad_norm": 0.11692175269126892,
"learning_rate": 0.000499746808903798,
"loss": 0.4103,
"step": 2850
},
{
"epoch": 0.16488514896520354,
"grad_norm": 0.11706849187612534,
"learning_rate": 0.0004997423669547419,
"loss": 0.4104,
"step": 2900
},
{
"epoch": 0.16772799636115535,
"grad_norm": 0.14066801965236664,
"learning_rate": 0.0004997379250056858,
"loss": 0.4104,
"step": 2950
},
{
"epoch": 0.17057084375710713,
"grad_norm": 0.14922069013118744,
"learning_rate": 0.0004997334830566296,
"loss": 0.4125,
"step": 3000
},
{
"epoch": 0.1734136911530589,
"grad_norm": 0.14433512091636658,
"learning_rate": 0.0004997290411075734,
"loss": 0.404,
"step": 3050
},
{
"epoch": 0.1762565385490107,
"grad_norm": 0.11257103830575943,
"learning_rate": 0.0004997245991585172,
"loss": 0.4089,
"step": 3100
},
{
"epoch": 0.17909938594496247,
"grad_norm": 0.13209494948387146,
"learning_rate": 0.000499720157209461,
"loss": 0.4081,
"step": 3150
},
{
"epoch": 0.18194223334091425,
"grad_norm": 0.13629132509231567,
"learning_rate": 0.0004997157152604048,
"loss": 0.4062,
"step": 3200
},
{
"epoch": 0.18478508073686606,
"grad_norm": 0.11546222865581512,
"learning_rate": 0.0004997112733113486,
"loss": 0.4037,
"step": 3250
},
{
"epoch": 0.18762792813281784,
"grad_norm": 0.12525054812431335,
"learning_rate": 0.0004997068313622925,
"loss": 0.4064,
"step": 3300
},
{
"epoch": 0.19047077552876962,
"grad_norm": 0.14944538474082947,
"learning_rate": 0.0004997023894132363,
"loss": 0.4085,
"step": 3350
},
{
"epoch": 0.1933136229247214,
"grad_norm": 0.15211829543113708,
"learning_rate": 0.0004996979474641801,
"loss": 0.4063,
"step": 3400
},
{
"epoch": 0.19615647032067318,
"grad_norm": 0.12800821661949158,
"learning_rate": 0.000499693505515124,
"loss": 0.4033,
"step": 3450
},
{
"epoch": 0.19899931771662496,
"grad_norm": 0.10901828110218048,
"learning_rate": 0.0004996890635660678,
"loss": 0.4041,
"step": 3500
},
{
"epoch": 0.20184216511257674,
"grad_norm": 0.13710814714431763,
"learning_rate": 0.0004996846216170116,
"loss": 0.4059,
"step": 3550
},
{
"epoch": 0.20468501250852855,
"grad_norm": 0.11688549071550369,
"learning_rate": 0.0004996801796679554,
"loss": 0.4019,
"step": 3600
},
{
"epoch": 0.20752785990448033,
"grad_norm": 0.10226283967494965,
"learning_rate": 0.0004996757377188992,
"loss": 0.4023,
"step": 3650
},
{
"epoch": 0.2103707073004321,
"grad_norm": 0.1010371595621109,
"learning_rate": 0.0004996712957698431,
"loss": 0.4031,
"step": 3700
},
{
"epoch": 0.2132135546963839,
"grad_norm": 0.12441287934780121,
"learning_rate": 0.0004996668538207869,
"loss": 0.4031,
"step": 3750
},
{
"epoch": 0.21605640209233568,
"grad_norm": 0.12629808485507965,
"learning_rate": 0.0004996624118717307,
"loss": 0.4046,
"step": 3800
},
{
"epoch": 0.21889924948828746,
"grad_norm": 0.1441943198442459,
"learning_rate": 0.0004996579699226745,
"loss": 0.4065,
"step": 3850
},
{
"epoch": 0.22174209688423926,
"grad_norm": 0.08717290312051773,
"learning_rate": 0.0004996535279736184,
"loss": 0.4043,
"step": 3900
},
{
"epoch": 0.22458494428019105,
"grad_norm": 0.12982551753520966,
"learning_rate": 0.0004996490860245623,
"loss": 0.3997,
"step": 3950
},
{
"epoch": 0.22742779167614283,
"grad_norm": 0.11208934336900711,
"learning_rate": 0.0004996446440755061,
"loss": 0.3966,
"step": 4000
},
{
"epoch": 0.2302706390720946,
"grad_norm": 0.13264520466327667,
"learning_rate": 0.0004996402021264499,
"loss": 0.401,
"step": 4050
},
{
"epoch": 0.2331134864680464,
"grad_norm": 0.1360943764448166,
"learning_rate": 0.0004996357601773937,
"loss": 0.3999,
"step": 4100
},
{
"epoch": 0.23595633386399817,
"grad_norm": 0.1145951896905899,
"learning_rate": 0.0004996313182283375,
"loss": 0.3995,
"step": 4150
},
{
"epoch": 0.23879918125994998,
"grad_norm": 0.12475068867206573,
"learning_rate": 0.0004996268762792813,
"loss": 0.4017,
"step": 4200
},
{
"epoch": 0.24164202865590176,
"grad_norm": 0.09585878252983093,
"learning_rate": 0.0004996224343302251,
"loss": 0.4007,
"step": 4250
},
{
"epoch": 0.24448487605185354,
"grad_norm": 0.11792745441198349,
"learning_rate": 0.000499617992381169,
"loss": 0.3991,
"step": 4300
},
{
"epoch": 0.24732772344780532,
"grad_norm": 0.16740038990974426,
"learning_rate": 0.0004996135504321129,
"loss": 0.3984,
"step": 4350
},
{
"epoch": 0.2501705708437571,
"grad_norm": 0.09640038013458252,
"learning_rate": 0.0004996091084830567,
"loss": 0.3976,
"step": 4400
},
{
"epoch": 0.2530134182397089,
"grad_norm": 0.09332569688558578,
"learning_rate": 0.0004996046665340005,
"loss": 0.3998,
"step": 4450
},
{
"epoch": 0.25585626563566066,
"grad_norm": 0.11328030377626419,
"learning_rate": 0.0004996002245849443,
"loss": 0.3979,
"step": 4500
},
{
"epoch": 0.25869911303161247,
"grad_norm": 0.1477869302034378,
"learning_rate": 0.0004995957826358881,
"loss": 0.3973,
"step": 4550
},
{
"epoch": 0.2615419604275642,
"grad_norm": 0.09623896330595016,
"learning_rate": 0.0004995913406868319,
"loss": 0.3962,
"step": 4600
},
{
"epoch": 0.26438480782351603,
"grad_norm": 0.11626730114221573,
"learning_rate": 0.0004995868987377757,
"loss": 0.3958,
"step": 4650
},
{
"epoch": 0.26722765521946784,
"grad_norm": 0.10988286882638931,
"learning_rate": 0.0004995824567887196,
"loss": 0.3944,
"step": 4700
},
{
"epoch": 0.2700705026154196,
"grad_norm": 0.12107770144939423,
"learning_rate": 0.0004995780148396634,
"loss": 0.3931,
"step": 4750
},
{
"epoch": 0.2729133500113714,
"grad_norm": 0.11602004617452621,
"learning_rate": 0.0004995735728906072,
"loss": 0.3963,
"step": 4800
},
{
"epoch": 0.27575619740732316,
"grad_norm": 0.10920259356498718,
"learning_rate": 0.0004995691309415511,
"loss": 0.395,
"step": 4850
},
{
"epoch": 0.27859904480327496,
"grad_norm": 0.1066461130976677,
"learning_rate": 0.0004995646889924949,
"loss": 0.3938,
"step": 4900
},
{
"epoch": 0.2814418921992268,
"grad_norm": 0.1071213036775589,
"learning_rate": 0.0004995602470434387,
"loss": 0.3943,
"step": 4950
},
{
"epoch": 0.2842847395951785,
"grad_norm": 0.08286375552415848,
"learning_rate": 0.0004995558050943825,
"loss": 0.3948,
"step": 5000
},
{
"epoch": 0.2842847395951785,
"eval_loss": 0.38135138154029846,
"eval_runtime": 583.2891,
"eval_samples_per_second": 226.149,
"eval_steps_per_second": 28.269,
"step": 5000
},
{
"epoch": 0.28712758699113033,
"grad_norm": 0.11819513142108917,
"learning_rate": 0.0004995513631453263,
"loss": 0.3969,
"step": 5050
},
{
"epoch": 0.2899704343870821,
"grad_norm": 0.10034069418907166,
"learning_rate": 0.0004995469211962702,
"loss": 0.3962,
"step": 5100
},
{
"epoch": 0.2928132817830339,
"grad_norm": 0.09750229120254517,
"learning_rate": 0.000499542479247214,
"loss": 0.3938,
"step": 5150
},
{
"epoch": 0.29565612917898565,
"grad_norm": 0.14667393267154694,
"learning_rate": 0.0004995380372981578,
"loss": 0.395,
"step": 5200
},
{
"epoch": 0.29849897657493746,
"grad_norm": 0.1200270876288414,
"learning_rate": 0.0004995335953491016,
"loss": 0.3939,
"step": 5250
},
{
"epoch": 0.30134182397088927,
"grad_norm": 0.12305665761232376,
"learning_rate": 0.0004995291534000456,
"loss": 0.3885,
"step": 5300
},
{
"epoch": 0.304184671366841,
"grad_norm": 0.09782176464796066,
"learning_rate": 0.0004995247114509894,
"loss": 0.3907,
"step": 5350
},
{
"epoch": 0.30702751876279283,
"grad_norm": 0.10045728832483292,
"learning_rate": 0.0004995202695019332,
"loss": 0.3921,
"step": 5400
},
{
"epoch": 0.3098703661587446,
"grad_norm": 0.15331445634365082,
"learning_rate": 0.000499515827552877,
"loss": 0.3953,
"step": 5450
},
{
"epoch": 0.3127132135546964,
"grad_norm": 0.10946525633335114,
"learning_rate": 0.0004995113856038208,
"loss": 0.3917,
"step": 5500
},
{
"epoch": 0.31555606095064814,
"grad_norm": 0.12001294642686844,
"learning_rate": 0.0004995069436547646,
"loss": 0.3915,
"step": 5550
},
{
"epoch": 0.31839890834659995,
"grad_norm": 0.11100505292415619,
"learning_rate": 0.0004995025017057084,
"loss": 0.3916,
"step": 5600
},
{
"epoch": 0.32124175574255176,
"grad_norm": 0.11148565262556076,
"learning_rate": 0.0004994980597566523,
"loss": 0.3947,
"step": 5650
},
{
"epoch": 0.3240846031385035,
"grad_norm": 0.10728944838047028,
"learning_rate": 0.0004994936178075961,
"loss": 0.3898,
"step": 5700
},
{
"epoch": 0.3269274505344553,
"grad_norm": 0.09397928416728973,
"learning_rate": 0.00049948917585854,
"loss": 0.3877,
"step": 5750
},
{
"epoch": 0.3297702979304071,
"grad_norm": 0.12607477605342865,
"learning_rate": 0.0004994847339094838,
"loss": 0.3919,
"step": 5800
},
{
"epoch": 0.3326131453263589,
"grad_norm": 0.11440327763557434,
"learning_rate": 0.0004994802919604276,
"loss": 0.3928,
"step": 5850
},
{
"epoch": 0.3354559927223107,
"grad_norm": 0.1275177150964737,
"learning_rate": 0.0004994758500113714,
"loss": 0.389,
"step": 5900
},
{
"epoch": 0.33829884011826244,
"grad_norm": 0.11246796697378159,
"learning_rate": 0.0004994714080623152,
"loss": 0.3862,
"step": 5950
},
{
"epoch": 0.34114168751421425,
"grad_norm": 0.11354568600654602,
"learning_rate": 0.000499466966113259,
"loss": 0.3908,
"step": 6000
},
{
"epoch": 0.343984534910166,
"grad_norm": 0.10693259537220001,
"learning_rate": 0.0004994625241642029,
"loss": 0.389,
"step": 6050
},
{
"epoch": 0.3468273823061178,
"grad_norm": 0.12362982332706451,
"learning_rate": 0.0004994580822151467,
"loss": 0.39,
"step": 6100
},
{
"epoch": 0.34967022970206957,
"grad_norm": 0.1343277543783188,
"learning_rate": 0.0004994536402660905,
"loss": 0.3934,
"step": 6150
},
{
"epoch": 0.3525130770980214,
"grad_norm": 0.15133637189865112,
"learning_rate": 0.0004994491983170344,
"loss": 0.382,
"step": 6200
},
{
"epoch": 0.3553559244939732,
"grad_norm": 0.11031973361968994,
"learning_rate": 0.0004994447563679782,
"loss": 0.3908,
"step": 6250
},
{
"epoch": 0.35819877188992494,
"grad_norm": 0.09201149642467499,
"learning_rate": 0.000499440314418922,
"loss": 0.3888,
"step": 6300
},
{
"epoch": 0.36104161928587675,
"grad_norm": 0.10396040976047516,
"learning_rate": 0.0004994358724698658,
"loss": 0.3861,
"step": 6350
},
{
"epoch": 0.3638844666818285,
"grad_norm": 0.11148199439048767,
"learning_rate": 0.0004994314305208096,
"loss": 0.3897,
"step": 6400
},
{
"epoch": 0.3667273140777803,
"grad_norm": 0.11479378491640091,
"learning_rate": 0.0004994269885717535,
"loss": 0.3883,
"step": 6450
},
{
"epoch": 0.3695701614737321,
"grad_norm": 0.09433585405349731,
"learning_rate": 0.0004994225466226973,
"loss": 0.3841,
"step": 6500
},
{
"epoch": 0.37241300886968387,
"grad_norm": 0.1304759830236435,
"learning_rate": 0.0004994181046736411,
"loss": 0.3876,
"step": 6550
},
{
"epoch": 0.3752558562656357,
"grad_norm": 0.11292573064565659,
"learning_rate": 0.0004994136627245849,
"loss": 0.3872,
"step": 6600
},
{
"epoch": 0.37809870366158743,
"grad_norm": 0.0928904116153717,
"learning_rate": 0.0004994092207755289,
"loss": 0.3875,
"step": 6650
},
{
"epoch": 0.38094155105753924,
"grad_norm": 0.10247938334941864,
"learning_rate": 0.0004994047788264727,
"loss": 0.3885,
"step": 6700
},
{
"epoch": 0.383784398453491,
"grad_norm": 0.11605235934257507,
"learning_rate": 0.0004994003368774165,
"loss": 0.3847,
"step": 6750
},
{
"epoch": 0.3866272458494428,
"grad_norm": 0.1496395617723465,
"learning_rate": 0.0004993958949283603,
"loss": 0.3861,
"step": 6800
},
{
"epoch": 0.3894700932453946,
"grad_norm": 0.11461085826158524,
"learning_rate": 0.0004993914529793041,
"loss": 0.3862,
"step": 6850
},
{
"epoch": 0.39231294064134636,
"grad_norm": 0.11239515244960785,
"learning_rate": 0.0004993870110302479,
"loss": 0.3865,
"step": 6900
},
{
"epoch": 0.3951557880372982,
"grad_norm": 0.11601614952087402,
"learning_rate": 0.0004993825690811917,
"loss": 0.3837,
"step": 6950
},
{
"epoch": 0.3979986354332499,
"grad_norm": 0.13389568030834198,
"learning_rate": 0.0004993781271321355,
"loss": 0.3866,
"step": 7000
},
{
"epoch": 0.40084148282920173,
"grad_norm": 0.12181925028562546,
"learning_rate": 0.0004993736851830794,
"loss": 0.388,
"step": 7050
},
{
"epoch": 0.4036843302251535,
"grad_norm": 0.10346251726150513,
"learning_rate": 0.0004993692432340233,
"loss": 0.3879,
"step": 7100
},
{
"epoch": 0.4065271776211053,
"grad_norm": 0.11786284297704697,
"learning_rate": 0.0004993648012849671,
"loss": 0.3859,
"step": 7150
},
{
"epoch": 0.4093700250170571,
"grad_norm": 0.10377379506826401,
"learning_rate": 0.0004993603593359109,
"loss": 0.3838,
"step": 7200
},
{
"epoch": 0.41221287241300886,
"grad_norm": 0.12437117844820023,
"learning_rate": 0.0004993559173868547,
"loss": 0.385,
"step": 7250
},
{
"epoch": 0.41505571980896067,
"grad_norm": 0.09871742129325867,
"learning_rate": 0.0004993514754377985,
"loss": 0.387,
"step": 7300
},
{
"epoch": 0.4178985672049124,
"grad_norm": 0.09648123383522034,
"learning_rate": 0.0004993470334887423,
"loss": 0.3843,
"step": 7350
},
{
"epoch": 0.4207414146008642,
"grad_norm": 0.14795717597007751,
"learning_rate": 0.0004993425915396861,
"loss": 0.3833,
"step": 7400
},
{
"epoch": 0.42358426199681604,
"grad_norm": 0.10902711004018784,
"learning_rate": 0.00049933814959063,
"loss": 0.3823,
"step": 7450
},
{
"epoch": 0.4264271093927678,
"grad_norm": 0.09183204919099808,
"learning_rate": 0.0004993337076415738,
"loss": 0.3871,
"step": 7500
},
{
"epoch": 0.4292699567887196,
"grad_norm": 0.11348464339971542,
"learning_rate": 0.0004993292656925176,
"loss": 0.3828,
"step": 7550
},
{
"epoch": 0.43211280418467135,
"grad_norm": 0.08863567560911179,
"learning_rate": 0.0004993248237434615,
"loss": 0.3817,
"step": 7600
},
{
"epoch": 0.43495565158062316,
"grad_norm": 0.1327882558107376,
"learning_rate": 0.0004993203817944053,
"loss": 0.3838,
"step": 7650
},
{
"epoch": 0.4377984989765749,
"grad_norm": 0.11749105900526047,
"learning_rate": 0.0004993159398453491,
"loss": 0.3865,
"step": 7700
},
{
"epoch": 0.4406413463725267,
"grad_norm": 0.09009282290935516,
"learning_rate": 0.0004993114978962929,
"loss": 0.3823,
"step": 7750
},
{
"epoch": 0.44348419376847853,
"grad_norm": 0.106484554708004,
"learning_rate": 0.0004993070559472367,
"loss": 0.3797,
"step": 7800
},
{
"epoch": 0.4463270411644303,
"grad_norm": 0.12862497568130493,
"learning_rate": 0.0004993026139981806,
"loss": 0.3804,
"step": 7850
},
{
"epoch": 0.4491698885603821,
"grad_norm": 0.1015087440609932,
"learning_rate": 0.0004992981720491244,
"loss": 0.3806,
"step": 7900
},
{
"epoch": 0.45201273595633384,
"grad_norm": 0.10782980173826218,
"learning_rate": 0.0004992937301000682,
"loss": 0.3858,
"step": 7950
},
{
"epoch": 0.45485558335228565,
"grad_norm": 0.09852251410484314,
"learning_rate": 0.000499289288151012,
"loss": 0.3824,
"step": 8000
},
{
"epoch": 0.45769843074823746,
"grad_norm": 0.10274305194616318,
"learning_rate": 0.000499284846201956,
"loss": 0.381,
"step": 8050
},
{
"epoch": 0.4605412781441892,
"grad_norm": 0.10695527493953705,
"learning_rate": 0.0004992804042528998,
"loss": 0.3839,
"step": 8100
},
{
"epoch": 0.463384125540141,
"grad_norm": 0.09151016920804977,
"learning_rate": 0.0004992759623038436,
"loss": 0.3842,
"step": 8150
},
{
"epoch": 0.4662269729360928,
"grad_norm": 0.15730440616607666,
"learning_rate": 0.0004992715203547874,
"loss": 0.3798,
"step": 8200
},
{
"epoch": 0.4690698203320446,
"grad_norm": 0.10243304073810577,
"learning_rate": 0.0004992670784057312,
"loss": 0.38,
"step": 8250
},
{
"epoch": 0.47191266772799634,
"grad_norm": 0.1174805536866188,
"learning_rate": 0.000499262636456675,
"loss": 0.3817,
"step": 8300
},
{
"epoch": 0.47475551512394815,
"grad_norm": 0.10942196846008301,
"learning_rate": 0.0004992581945076188,
"loss": 0.3829,
"step": 8350
},
{
"epoch": 0.47759836251989995,
"grad_norm": 0.10109516978263855,
"learning_rate": 0.0004992537525585626,
"loss": 0.3803,
"step": 8400
},
{
"epoch": 0.4804412099158517,
"grad_norm": 0.09248687326908112,
"learning_rate": 0.0004992493106095065,
"loss": 0.3796,
"step": 8450
},
{
"epoch": 0.4832840573118035,
"grad_norm": 0.11525557935237885,
"learning_rate": 0.0004992448686604504,
"loss": 0.3823,
"step": 8500
},
{
"epoch": 0.48612690470775527,
"grad_norm": 0.10481023788452148,
"learning_rate": 0.0004992404267113942,
"loss": 0.3781,
"step": 8550
},
{
"epoch": 0.4889697521037071,
"grad_norm": 0.09493274986743927,
"learning_rate": 0.000499235984762338,
"loss": 0.38,
"step": 8600
},
{
"epoch": 0.49181259949965883,
"grad_norm": 0.08859369158744812,
"learning_rate": 0.0004992315428132818,
"loss": 0.3789,
"step": 8650
},
{
"epoch": 0.49465544689561064,
"grad_norm": 0.11814738810062408,
"learning_rate": 0.0004992271008642256,
"loss": 0.3809,
"step": 8700
},
{
"epoch": 0.49749829429156245,
"grad_norm": 0.1322488635778427,
"learning_rate": 0.0004992226589151694,
"loss": 0.3795,
"step": 8750
},
{
"epoch": 0.5003411416875142,
"grad_norm": 0.1304454803466797,
"learning_rate": 0.0004992182169661132,
"loss": 0.3794,
"step": 8800
},
{
"epoch": 0.503183989083466,
"grad_norm": 0.09153052419424057,
"learning_rate": 0.0004992137750170571,
"loss": 0.3769,
"step": 8850
},
{
"epoch": 0.5060268364794178,
"grad_norm": 0.10608222335577011,
"learning_rate": 0.0004992093330680009,
"loss": 0.3748,
"step": 8900
},
{
"epoch": 0.5088696838753696,
"grad_norm": 0.1054573506116867,
"learning_rate": 0.0004992048911189447,
"loss": 0.3817,
"step": 8950
},
{
"epoch": 0.5117125312713213,
"grad_norm": 0.1694241166114807,
"learning_rate": 0.0004992004491698886,
"loss": 0.3774,
"step": 9000
},
{
"epoch": 0.5145553786672732,
"grad_norm": 0.13877439498901367,
"learning_rate": 0.0004991960072208324,
"loss": 0.3769,
"step": 9050
},
{
"epoch": 0.5173982260632249,
"grad_norm": 0.10788954794406891,
"learning_rate": 0.0004991915652717762,
"loss": 0.3819,
"step": 9100
},
{
"epoch": 0.5202410734591767,
"grad_norm": 0.12326483428478241,
"learning_rate": 0.00049918712332272,
"loss": 0.3812,
"step": 9150
},
{
"epoch": 0.5230839208551284,
"grad_norm": 0.07986534386873245,
"learning_rate": 0.0004991826813736638,
"loss": 0.3797,
"step": 9200
},
{
"epoch": 0.5259267682510803,
"grad_norm": 0.11102836579084396,
"learning_rate": 0.0004991782394246077,
"loss": 0.3797,
"step": 9250
},
{
"epoch": 0.5287696156470321,
"grad_norm": 0.10697384923696518,
"learning_rate": 0.0004991737974755515,
"loss": 0.3767,
"step": 9300
},
{
"epoch": 0.5316124630429838,
"grad_norm": 0.11679094284772873,
"learning_rate": 0.0004991693555264953,
"loss": 0.3782,
"step": 9350
},
{
"epoch": 0.5344553104389357,
"grad_norm": 0.10827042162418365,
"learning_rate": 0.0004991649135774391,
"loss": 0.376,
"step": 9400
},
{
"epoch": 0.5372981578348874,
"grad_norm": 0.0934649109840393,
"learning_rate": 0.0004991604716283831,
"loss": 0.3775,
"step": 9450
},
{
"epoch": 0.5401410052308392,
"grad_norm": 0.08999691903591156,
"learning_rate": 0.0004991560296793269,
"loss": 0.3783,
"step": 9500
},
{
"epoch": 0.5429838526267909,
"grad_norm": 0.11615636944770813,
"learning_rate": 0.0004991515877302707,
"loss": 0.3779,
"step": 9550
},
{
"epoch": 0.5458267000227428,
"grad_norm": 0.4005848467350006,
"learning_rate": 0.0004991471457812145,
"loss": 0.3745,
"step": 9600
},
{
"epoch": 0.5486695474186946,
"grad_norm": 0.12695498764514923,
"learning_rate": 0.0004991427038321583,
"loss": 0.3916,
"step": 9650
},
{
"epoch": 0.5515123948146463,
"grad_norm": 0.10009093582630157,
"learning_rate": 0.0004991382618831021,
"loss": 0.3798,
"step": 9700
},
{
"epoch": 0.5543552422105982,
"grad_norm": 0.1480248123407364,
"learning_rate": 0.0004991338199340459,
"loss": 0.3752,
"step": 9750
},
{
"epoch": 0.5571980896065499,
"grad_norm": 0.10428158193826675,
"learning_rate": 0.0004991293779849897,
"loss": 0.3787,
"step": 9800
},
{
"epoch": 0.5600409370025017,
"grad_norm": 0.1932564377784729,
"learning_rate": 0.0004991249360359336,
"loss": 0.3812,
"step": 9850
},
{
"epoch": 0.5628837843984535,
"grad_norm": 0.088712178170681,
"learning_rate": 0.0004991204940868775,
"loss": 0.3783,
"step": 9900
},
{
"epoch": 0.5657266317944053,
"grad_norm": 0.10896250605583191,
"learning_rate": 0.0004991160521378213,
"loss": 0.3782,
"step": 9950
},
{
"epoch": 0.568569479190357,
"grad_norm": 0.1077795922756195,
"learning_rate": 0.0004991116101887651,
"loss": 0.3773,
"step": 10000
},
{
"epoch": 0.568569479190357,
"eval_loss": 0.36588892340660095,
"eval_runtime": 584.439,
"eval_samples_per_second": 225.704,
"eval_steps_per_second": 28.213,
"step": 10000
},
{
"epoch": 0.5714123265863088,
"grad_norm": 0.09190759807825089,
"learning_rate": 0.0004991071682397089,
"loss": 0.3785,
"step": 10050
},
{
"epoch": 0.5742551739822607,
"grad_norm": 0.11082971841096878,
"learning_rate": 0.0004991027262906527,
"loss": 0.3775,
"step": 10100
},
{
"epoch": 0.5770980213782124,
"grad_norm": 0.10099225491285324,
"learning_rate": 0.0004990982843415965,
"loss": 0.3775,
"step": 10150
},
{
"epoch": 0.5799408687741642,
"grad_norm": 0.1112075224518776,
"learning_rate": 0.0004990938423925403,
"loss": 0.3764,
"step": 10200
},
{
"epoch": 0.582783716170116,
"grad_norm": 0.09865789860486984,
"learning_rate": 0.0004990894004434842,
"loss": 0.3802,
"step": 10250
},
{
"epoch": 0.5856265635660678,
"grad_norm": 0.10351789742708206,
"learning_rate": 0.000499084958494428,
"loss": 0.3795,
"step": 10300
},
{
"epoch": 0.5884694109620195,
"grad_norm": 0.10376661270856857,
"learning_rate": 0.0004990805165453719,
"loss": 0.3753,
"step": 10350
},
{
"epoch": 0.5913122583579713,
"grad_norm": 0.09663544595241547,
"learning_rate": 0.0004990760745963157,
"loss": 0.3737,
"step": 10400
},
{
"epoch": 0.5941551057539232,
"grad_norm": 0.09701116383075714,
"learning_rate": 0.0004990716326472595,
"loss": 0.3737,
"step": 10450
},
{
"epoch": 0.5969979531498749,
"grad_norm": 0.1021418496966362,
"learning_rate": 0.0004990671906982033,
"loss": 0.3728,
"step": 10500
},
{
"epoch": 0.5998408005458267,
"grad_norm": 0.12170026451349258,
"learning_rate": 0.0004990627487491472,
"loss": 0.3755,
"step": 10550
},
{
"epoch": 0.6026836479417785,
"grad_norm": 0.1142646074295044,
"learning_rate": 0.000499058306800091,
"loss": 0.378,
"step": 10600
},
{
"epoch": 0.6055264953377303,
"grad_norm": 0.09568295627832413,
"learning_rate": 0.0004990538648510348,
"loss": 0.3742,
"step": 10650
},
{
"epoch": 0.608369342733682,
"grad_norm": 0.11586258560419083,
"learning_rate": 0.0004990494229019786,
"loss": 0.3731,
"step": 10700
},
{
"epoch": 0.6112121901296338,
"grad_norm": 0.0984320268034935,
"learning_rate": 0.0004990449809529224,
"loss": 0.3767,
"step": 10750
},
{
"epoch": 0.6140550375255857,
"grad_norm": 0.10665503889322281,
"learning_rate": 0.0004990405390038662,
"loss": 0.3756,
"step": 10800
},
{
"epoch": 0.6168978849215374,
"grad_norm": 0.10374121367931366,
"learning_rate": 0.0004990360970548102,
"loss": 0.376,
"step": 10850
},
{
"epoch": 0.6197407323174892,
"grad_norm": 0.10508013516664505,
"learning_rate": 0.000499031655105754,
"loss": 0.3717,
"step": 10900
},
{
"epoch": 0.622583579713441,
"grad_norm": 0.10919308662414551,
"learning_rate": 0.0004990272131566978,
"loss": 0.3746,
"step": 10950
},
{
"epoch": 0.6254264271093928,
"grad_norm": 0.09990997612476349,
"learning_rate": 0.0004990227712076416,
"loss": 0.3738,
"step": 11000
},
{
"epoch": 0.6282692745053445,
"grad_norm": 0.10817346721887589,
"learning_rate": 0.0004990183292585854,
"loss": 0.3688,
"step": 11050
},
{
"epoch": 0.6311121219012963,
"grad_norm": 0.0922752320766449,
"learning_rate": 0.0004990138873095292,
"loss": 0.3751,
"step": 11100
},
{
"epoch": 0.6339549692972481,
"grad_norm": 0.09761642664670944,
"learning_rate": 0.000499009445360473,
"loss": 0.3724,
"step": 11150
},
{
"epoch": 0.6367978166931999,
"grad_norm": 0.09401151537895203,
"learning_rate": 0.0004990050034114168,
"loss": 0.3717,
"step": 11200
},
{
"epoch": 0.6396406640891517,
"grad_norm": 0.10084612667560577,
"learning_rate": 0.0004990005614623607,
"loss": 0.3711,
"step": 11250
},
{
"epoch": 0.6424835114851035,
"grad_norm": 0.11709214001893997,
"learning_rate": 0.0004989961195133046,
"loss": 0.3739,
"step": 11300
},
{
"epoch": 0.6453263588810553,
"grad_norm": 0.1035287156701088,
"learning_rate": 0.0004989916775642484,
"loss": 0.3684,
"step": 11350
},
{
"epoch": 0.648169206277007,
"grad_norm": 0.09968025982379913,
"learning_rate": 0.0004989872356151922,
"loss": 0.3749,
"step": 11400
},
{
"epoch": 0.6510120536729589,
"grad_norm": 0.1003747433423996,
"learning_rate": 0.000498982793666136,
"loss": 0.3721,
"step": 11450
},
{
"epoch": 0.6538549010689106,
"grad_norm": 0.09462849795818329,
"learning_rate": 0.0004989783517170798,
"loss": 0.3687,
"step": 11500
},
{
"epoch": 0.6566977484648624,
"grad_norm": 0.10089970380067825,
"learning_rate": 0.0004989739097680236,
"loss": 0.3724,
"step": 11550
},
{
"epoch": 0.6595405958608141,
"grad_norm": 0.11282893270254135,
"learning_rate": 0.0004989694678189675,
"loss": 0.3726,
"step": 11600
},
{
"epoch": 0.662383443256766,
"grad_norm": 0.12321527302265167,
"learning_rate": 0.0004989650258699113,
"loss": 0.3704,
"step": 11650
},
{
"epoch": 0.6652262906527178,
"grad_norm": 0.10225247591733932,
"learning_rate": 0.0004989605839208551,
"loss": 0.3717,
"step": 11700
},
{
"epoch": 0.6680691380486695,
"grad_norm": 0.09634348750114441,
"learning_rate": 0.000498956141971799,
"loss": 0.3731,
"step": 11750
},
{
"epoch": 0.6709119854446214,
"grad_norm": 0.09670031815767288,
"learning_rate": 0.0004989517000227428,
"loss": 0.3697,
"step": 11800
},
{
"epoch": 0.6737548328405731,
"grad_norm": 0.11874838918447495,
"learning_rate": 0.0004989472580736867,
"loss": 0.3717,
"step": 11850
},
{
"epoch": 0.6765976802365249,
"grad_norm": 0.1474309116601944,
"learning_rate": 0.0004989428161246305,
"loss": 0.3719,
"step": 11900
},
{
"epoch": 0.6794405276324766,
"grad_norm": 0.0953158363699913,
"learning_rate": 0.0004989383741755743,
"loss": 0.3698,
"step": 11950
},
{
"epoch": 0.6822833750284285,
"grad_norm": 0.07613354921340942,
"learning_rate": 0.0004989339322265181,
"loss": 0.372,
"step": 12000
},
{
"epoch": 0.6851262224243803,
"grad_norm": 0.10099935531616211,
"learning_rate": 0.0004989294902774619,
"loss": 0.3715,
"step": 12050
},
{
"epoch": 0.687969069820332,
"grad_norm": 0.1142035722732544,
"learning_rate": 0.0004989250483284057,
"loss": 0.371,
"step": 12100
},
{
"epoch": 0.6908119172162839,
"grad_norm": 0.11868602782487869,
"learning_rate": 0.0004989206063793495,
"loss": 0.3718,
"step": 12150
},
{
"epoch": 0.6936547646122356,
"grad_norm": 0.11403531581163406,
"learning_rate": 0.0004989161644302935,
"loss": 0.37,
"step": 12200
},
{
"epoch": 0.6964976120081874,
"grad_norm": 0.10048527270555496,
"learning_rate": 0.0004989117224812373,
"loss": 0.3689,
"step": 12250
},
{
"epoch": 0.6993404594041391,
"grad_norm": 0.0894421935081482,
"learning_rate": 0.0004989072805321811,
"loss": 0.3675,
"step": 12300
},
{
"epoch": 0.702183306800091,
"grad_norm": 0.10465215891599655,
"learning_rate": 0.0004989028385831249,
"loss": 0.3707,
"step": 12350
},
{
"epoch": 0.7050261541960428,
"grad_norm": 0.10983111709356308,
"learning_rate": 0.0004988983966340687,
"loss": 0.3713,
"step": 12400
},
{
"epoch": 0.7078690015919945,
"grad_norm": 0.11033093929290771,
"learning_rate": 0.0004988939546850125,
"loss": 0.3723,
"step": 12450
},
{
"epoch": 0.7107118489879464,
"grad_norm": 0.11053282022476196,
"learning_rate": 0.0004988895127359563,
"loss": 0.3701,
"step": 12500
},
{
"epoch": 0.7135546963838981,
"grad_norm": 0.09599506109952927,
"learning_rate": 0.0004988850707869001,
"loss": 0.3696,
"step": 12550
},
{
"epoch": 0.7163975437798499,
"grad_norm": 0.10971349477767944,
"learning_rate": 0.000498880628837844,
"loss": 0.3692,
"step": 12600
},
{
"epoch": 0.7192403911758016,
"grad_norm": 0.09890740364789963,
"learning_rate": 0.0004988761868887879,
"loss": 0.3671,
"step": 12650
},
{
"epoch": 0.7220832385717535,
"grad_norm": 0.09667664021253586,
"learning_rate": 0.0004988717449397317,
"loss": 0.3708,
"step": 12700
},
{
"epoch": 0.7249260859677052,
"grad_norm": 0.08852635324001312,
"learning_rate": 0.0004988673029906755,
"loss": 0.3704,
"step": 12750
},
{
"epoch": 0.727768933363657,
"grad_norm": 0.08930686861276627,
"learning_rate": 0.0004988628610416193,
"loss": 0.3676,
"step": 12800
},
{
"epoch": 0.7306117807596089,
"grad_norm": 0.10307100415229797,
"learning_rate": 0.0004988584190925631,
"loss": 0.3702,
"step": 12850
},
{
"epoch": 0.7334546281555606,
"grad_norm": 0.09995568543672562,
"learning_rate": 0.0004988539771435069,
"loss": 0.367,
"step": 12900
},
{
"epoch": 0.7362974755515124,
"grad_norm": 0.09159702807664871,
"learning_rate": 0.0004988495351944507,
"loss": 0.3661,
"step": 12950
},
{
"epoch": 0.7391403229474642,
"grad_norm": 0.0963875874876976,
"learning_rate": 0.0004988450932453946,
"loss": 0.3667,
"step": 13000
},
{
"epoch": 0.741983170343416,
"grad_norm": 0.09584073722362518,
"learning_rate": 0.0004988406512963384,
"loss": 0.3697,
"step": 13050
},
{
"epoch": 0.7448260177393677,
"grad_norm": 0.09499222785234451,
"learning_rate": 0.0004988362093472822,
"loss": 0.3692,
"step": 13100
},
{
"epoch": 0.7476688651353195,
"grad_norm": 0.10076665878295898,
"learning_rate": 0.0004988317673982261,
"loss": 0.3672,
"step": 13150
},
{
"epoch": 0.7505117125312714,
"grad_norm": 0.1046692505478859,
"learning_rate": 0.00049882732544917,
"loss": 0.3669,
"step": 13200
},
{
"epoch": 0.7533545599272231,
"grad_norm": 0.11345544457435608,
"learning_rate": 0.0004988228835001138,
"loss": 0.3676,
"step": 13250
},
{
"epoch": 0.7561974073231749,
"grad_norm": 0.09689969569444656,
"learning_rate": 0.0004988184415510576,
"loss": 0.367,
"step": 13300
},
{
"epoch": 0.7590402547191267,
"grad_norm": 0.10489045083522797,
"learning_rate": 0.0004988139996020014,
"loss": 0.3681,
"step": 13350
},
{
"epoch": 0.7618831021150785,
"grad_norm": 0.09123999625444412,
"learning_rate": 0.0004988095576529452,
"loss": 0.3687,
"step": 13400
},
{
"epoch": 0.7647259495110302,
"grad_norm": 0.09405972063541412,
"learning_rate": 0.000498805115703889,
"loss": 0.3715,
"step": 13450
},
{
"epoch": 0.767568796906982,
"grad_norm": 0.08794659376144409,
"learning_rate": 0.0004988006737548328,
"loss": 0.3653,
"step": 13500
},
{
"epoch": 0.7704116443029339,
"grad_norm": 0.11547774076461792,
"learning_rate": 0.0004987962318057766,
"loss": 0.3677,
"step": 13550
},
{
"epoch": 0.7732544916988856,
"grad_norm": 0.0955629050731659,
"learning_rate": 0.0004987917898567206,
"loss": 0.369,
"step": 13600
},
{
"epoch": 0.7760973390948374,
"grad_norm": 0.10582061856985092,
"learning_rate": 0.0004987873479076644,
"loss": 0.3696,
"step": 13650
},
{
"epoch": 0.7789401864907892,
"grad_norm": 0.11626145988702774,
"learning_rate": 0.0004987829059586082,
"loss": 0.369,
"step": 13700
},
{
"epoch": 0.781783033886741,
"grad_norm": 0.08376283198595047,
"learning_rate": 0.000498778464009552,
"loss": 0.3686,
"step": 13750
},
{
"epoch": 0.7846258812826927,
"grad_norm": 0.09740207344293594,
"learning_rate": 0.0004987740220604958,
"loss": 0.3649,
"step": 13800
},
{
"epoch": 0.7874687286786445,
"grad_norm": 0.08660556375980377,
"learning_rate": 0.0004987695801114396,
"loss": 0.3654,
"step": 13850
},
{
"epoch": 0.7903115760745963,
"grad_norm": 0.09063564240932465,
"learning_rate": 0.0004987651381623834,
"loss": 0.3671,
"step": 13900
},
{
"epoch": 0.7931544234705481,
"grad_norm": 0.12205325812101364,
"learning_rate": 0.0004987606962133272,
"loss": 0.3667,
"step": 13950
},
{
"epoch": 0.7959972708664999,
"grad_norm": 0.1259283423423767,
"learning_rate": 0.0004987562542642711,
"loss": 0.3611,
"step": 14000
},
{
"epoch": 0.7988401182624517,
"grad_norm": 0.0938865914940834,
"learning_rate": 0.000498751812315215,
"loss": 0.3666,
"step": 14050
},
{
"epoch": 0.8016829656584035,
"grad_norm": 0.10185267776250839,
"learning_rate": 0.0004987473703661588,
"loss": 0.3648,
"step": 14100
},
{
"epoch": 0.8045258130543552,
"grad_norm": 0.11661996692419052,
"learning_rate": 0.0004987429284171026,
"loss": 0.3657,
"step": 14150
},
{
"epoch": 0.807368660450307,
"grad_norm": 0.0994558110833168,
"learning_rate": 0.0004987384864680464,
"loss": 0.3639,
"step": 14200
},
{
"epoch": 0.8102115078462588,
"grad_norm": 0.09819088876247406,
"learning_rate": 0.0004987340445189902,
"loss": 0.3642,
"step": 14250
},
{
"epoch": 0.8130543552422106,
"grad_norm": 0.09436295926570892,
"learning_rate": 0.000498729602569934,
"loss": 0.3677,
"step": 14300
},
{
"epoch": 0.8158972026381623,
"grad_norm": 0.12002996355295181,
"learning_rate": 0.0004987251606208778,
"loss": 0.366,
"step": 14350
},
{
"epoch": 0.8187400500341142,
"grad_norm": 0.08650200068950653,
"learning_rate": 0.0004987207186718217,
"loss": 0.3645,
"step": 14400
},
{
"epoch": 0.821582897430066,
"grad_norm": 0.1826743632555008,
"learning_rate": 0.0004987162767227655,
"loss": 0.3681,
"step": 14450
},
{
"epoch": 0.8244257448260177,
"grad_norm": 0.10164210200309753,
"learning_rate": 0.0004987118347737093,
"loss": 0.3706,
"step": 14500
},
{
"epoch": 0.8272685922219696,
"grad_norm": 0.09652096033096313,
"learning_rate": 0.0004987073928246532,
"loss": 0.364,
"step": 14550
},
{
"epoch": 0.8301114396179213,
"grad_norm": 0.1318165808916092,
"learning_rate": 0.0004987029508755971,
"loss": 0.3651,
"step": 14600
},
{
"epoch": 0.8329542870138731,
"grad_norm": 0.09845109283924103,
"learning_rate": 0.0004986985089265409,
"loss": 0.3636,
"step": 14650
},
{
"epoch": 0.8357971344098248,
"grad_norm": 0.12955370545387268,
"learning_rate": 0.0004986940669774847,
"loss": 0.3641,
"step": 14700
},
{
"epoch": 0.8386399818057767,
"grad_norm": 0.10223904252052307,
"learning_rate": 0.0004986896250284285,
"loss": 0.363,
"step": 14750
},
{
"epoch": 0.8414828292017285,
"grad_norm": 0.09701918810606003,
"learning_rate": 0.0004986851830793723,
"loss": 0.3627,
"step": 14800
},
{
"epoch": 0.8443256765976802,
"grad_norm": 0.0875435471534729,
"learning_rate": 0.0004986807411303161,
"loss": 0.3651,
"step": 14850
},
{
"epoch": 0.8471685239936321,
"grad_norm": 0.09006401896476746,
"learning_rate": 0.0004986762991812599,
"loss": 0.3632,
"step": 14900
},
{
"epoch": 0.8500113713895838,
"grad_norm": 0.09841691702604294,
"learning_rate": 0.0004986718572322037,
"loss": 0.3641,
"step": 14950
},
{
"epoch": 0.8528542187855356,
"grad_norm": 0.11014077812433243,
"learning_rate": 0.0004986674152831477,
"loss": 0.3654,
"step": 15000
},
{
"epoch": 0.8528542187855356,
"eval_loss": 0.3545059561729431,
"eval_runtime": 572.3313,
"eval_samples_per_second": 230.478,
"eval_steps_per_second": 28.81,
"step": 15000
},
{
"epoch": 0.8556970661814873,
"grad_norm": 0.11560770869255066,
"learning_rate": 0.0004986629733340915,
"loss": 0.3644,
"step": 15050
},
{
"epoch": 0.8585399135774392,
"grad_norm": 0.08875507861375809,
"learning_rate": 0.0004986585313850353,
"loss": 0.3637,
"step": 15100
},
{
"epoch": 0.861382760973391,
"grad_norm": 0.09458891302347183,
"learning_rate": 0.0004986540894359791,
"loss": 0.3638,
"step": 15150
},
{
"epoch": 0.8642256083693427,
"grad_norm": 0.10495459288358688,
"learning_rate": 0.0004986496474869229,
"loss": 0.3647,
"step": 15200
},
{
"epoch": 0.8670684557652946,
"grad_norm": 0.10532251745462418,
"learning_rate": 0.0004986452055378667,
"loss": 0.3664,
"step": 15250
},
{
"epoch": 0.8699113031612463,
"grad_norm": 0.09231790155172348,
"learning_rate": 0.0004986407635888105,
"loss": 0.3643,
"step": 15300
},
{
"epoch": 0.8727541505571981,
"grad_norm": 0.09325330704450607,
"learning_rate": 0.0004986363216397543,
"loss": 0.3659,
"step": 15350
},
{
"epoch": 0.8755969979531498,
"grad_norm": 0.11519546806812286,
"learning_rate": 0.0004986318796906982,
"loss": 0.3654,
"step": 15400
},
{
"epoch": 0.8784398453491017,
"grad_norm": 0.08858964592218399,
"learning_rate": 0.0004986274377416421,
"loss": 0.3647,
"step": 15450
},
{
"epoch": 0.8812826927450534,
"grad_norm": 0.11211369931697845,
"learning_rate": 0.0004986229957925859,
"loss": 0.3646,
"step": 15500
},
{
"epoch": 0.8841255401410052,
"grad_norm": 0.09691707044839859,
"learning_rate": 0.0004986185538435297,
"loss": 0.3633,
"step": 15550
},
{
"epoch": 0.8869683875369571,
"grad_norm": 0.09831210970878601,
"learning_rate": 0.0004986141118944735,
"loss": 0.3644,
"step": 15600
},
{
"epoch": 0.8898112349329088,
"grad_norm": 0.10148192197084427,
"learning_rate": 0.0004986096699454173,
"loss": 0.3613,
"step": 15650
},
{
"epoch": 0.8926540823288606,
"grad_norm": 0.0974993035197258,
"learning_rate": 0.0004986052279963611,
"loss": 0.3648,
"step": 15700
},
{
"epoch": 0.8954969297248123,
"grad_norm": 0.09062571078538895,
"learning_rate": 0.0004986007860473049,
"loss": 0.3622,
"step": 15750
},
{
"epoch": 0.8983397771207642,
"grad_norm": 0.11505638062953949,
"learning_rate": 0.0004985963440982488,
"loss": 0.3647,
"step": 15800
},
{
"epoch": 0.9011826245167159,
"grad_norm": 0.10409519821405411,
"learning_rate": 0.0004985919021491926,
"loss": 0.3684,
"step": 15850
},
{
"epoch": 0.9040254719126677,
"grad_norm": 0.09695091098546982,
"learning_rate": 0.0004985874602001365,
"loss": 0.3627,
"step": 15900
},
{
"epoch": 0.9068683193086196,
"grad_norm": 0.10802540183067322,
"learning_rate": 0.0004985830182510803,
"loss": 0.363,
"step": 15950
},
{
"epoch": 0.9097111667045713,
"grad_norm": 0.08416867256164551,
"learning_rate": 0.0004985785763020242,
"loss": 0.3633,
"step": 16000
},
{
"epoch": 0.9125540141005231,
"grad_norm": 0.12350185215473175,
"learning_rate": 0.000498574134352968,
"loss": 0.3645,
"step": 16050
},
{
"epoch": 0.9153968614964749,
"grad_norm": 0.10537421703338623,
"learning_rate": 0.0004985696924039118,
"loss": 0.3635,
"step": 16100
},
{
"epoch": 0.9182397088924267,
"grad_norm": 0.09744290262460709,
"learning_rate": 0.0004985652504548556,
"loss": 0.3623,
"step": 16150
},
{
"epoch": 0.9210825562883784,
"grad_norm": 0.09753882884979248,
"learning_rate": 0.0004985608085057994,
"loss": 0.3617,
"step": 16200
},
{
"epoch": 0.9239254036843302,
"grad_norm": 0.08662727475166321,
"learning_rate": 0.0004985563665567432,
"loss": 0.3624,
"step": 16250
},
{
"epoch": 0.926768251080282,
"grad_norm": 0.09310087561607361,
"learning_rate": 0.000498551924607687,
"loss": 0.3631,
"step": 16300
},
{
"epoch": 0.9296110984762338,
"grad_norm": 0.10131137073040009,
"learning_rate": 0.0004985474826586308,
"loss": 0.3593,
"step": 16350
},
{
"epoch": 0.9324539458721856,
"grad_norm": 0.09597592055797577,
"learning_rate": 0.0004985430407095748,
"loss": 0.3628,
"step": 16400
},
{
"epoch": 0.9352967932681374,
"grad_norm": 0.09163256734609604,
"learning_rate": 0.0004985385987605186,
"loss": 0.3596,
"step": 16450
},
{
"epoch": 0.9381396406640892,
"grad_norm": 0.11283744126558304,
"learning_rate": 0.0004985341568114624,
"loss": 0.3629,
"step": 16500
},
{
"epoch": 0.9409824880600409,
"grad_norm": 0.1027892455458641,
"learning_rate": 0.0004985297148624062,
"loss": 0.3624,
"step": 16550
},
{
"epoch": 0.9438253354559927,
"grad_norm": 0.09853541105985641,
"learning_rate": 0.00049852527291335,
"loss": 0.3606,
"step": 16600
},
{
"epoch": 0.9466681828519445,
"grad_norm": 0.09280608594417572,
"learning_rate": 0.0004985208309642938,
"loss": 0.3626,
"step": 16650
},
{
"epoch": 0.9495110302478963,
"grad_norm": 0.09166574478149414,
"learning_rate": 0.0004985163890152376,
"loss": 0.3631,
"step": 16700
},
{
"epoch": 0.952353877643848,
"grad_norm": 0.10630334913730621,
"learning_rate": 0.0004985119470661814,
"loss": 0.3611,
"step": 16750
},
{
"epoch": 0.9551967250397999,
"grad_norm": 0.09024935215711594,
"learning_rate": 0.0004985075051171253,
"loss": 0.3627,
"step": 16800
},
{
"epoch": 0.9580395724357517,
"grad_norm": 0.08191289007663727,
"learning_rate": 0.0004985030631680692,
"loss": 0.3586,
"step": 16850
},
{
"epoch": 0.9608824198317034,
"grad_norm": 0.08423138409852982,
"learning_rate": 0.000498498621219013,
"loss": 0.3622,
"step": 16900
},
{
"epoch": 0.9637252672276552,
"grad_norm": 0.09373466670513153,
"learning_rate": 0.0004984941792699568,
"loss": 0.3611,
"step": 16950
},
{
"epoch": 0.966568114623607,
"grad_norm": 0.09812022745609283,
"learning_rate": 0.0004984897373209006,
"loss": 0.3627,
"step": 17000
},
{
"epoch": 0.9694109620195588,
"grad_norm": 0.09764017909765244,
"learning_rate": 0.0004984852953718444,
"loss": 0.3589,
"step": 17050
},
{
"epoch": 0.9722538094155105,
"grad_norm": 0.10176458954811096,
"learning_rate": 0.0004984808534227883,
"loss": 0.3594,
"step": 17100
},
{
"epoch": 0.9750966568114624,
"grad_norm": 0.10278456658124924,
"learning_rate": 0.0004984764114737321,
"loss": 0.3581,
"step": 17150
},
{
"epoch": 0.9779395042074142,
"grad_norm": 0.1218334510922432,
"learning_rate": 0.0004984719695246759,
"loss": 0.3606,
"step": 17200
},
{
"epoch": 0.9807823516033659,
"grad_norm": 0.09451757371425629,
"learning_rate": 0.0004984675275756197,
"loss": 0.3614,
"step": 17250
},
{
"epoch": 0.9836251989993177,
"grad_norm": 0.10088694840669632,
"learning_rate": 0.0004984630856265636,
"loss": 0.3604,
"step": 17300
},
{
"epoch": 0.9864680463952695,
"grad_norm": 0.10026043653488159,
"learning_rate": 0.0004984586436775075,
"loss": 0.3604,
"step": 17350
},
{
"epoch": 0.9893108937912213,
"grad_norm": 0.11412831395864487,
"learning_rate": 0.0004984542017284513,
"loss": 0.3608,
"step": 17400
},
{
"epoch": 0.992153741187173,
"grad_norm": 0.10250277817249298,
"learning_rate": 0.0004984497597793951,
"loss": 0.3597,
"step": 17450
},
{
"epoch": 0.9949965885831249,
"grad_norm": 0.10922574251890182,
"learning_rate": 0.0004984453178303389,
"loss": 0.3611,
"step": 17500
},
{
"epoch": 0.9978394359790766,
"grad_norm": 0.09554197639226913,
"learning_rate": 0.0004984408758812827,
"loss": 0.3596,
"step": 17550
},
{
"epoch": 1.0006822833750284,
"grad_norm": 0.11213108897209167,
"learning_rate": 0.0004984364339322265,
"loss": 0.3617,
"step": 17600
},
{
"epoch": 1.0035251307709803,
"grad_norm": 0.13735197484493256,
"learning_rate": 0.0004984319919831703,
"loss": 0.3528,
"step": 17650
},
{
"epoch": 1.006367978166932,
"grad_norm": 0.10477373749017715,
"learning_rate": 0.0004984275500341141,
"loss": 0.3561,
"step": 17700
},
{
"epoch": 1.0092108255628838,
"grad_norm": 0.10232077538967133,
"learning_rate": 0.0004984231080850581,
"loss": 0.3563,
"step": 17750
},
{
"epoch": 1.0120536729588356,
"grad_norm": 0.11949522793292999,
"learning_rate": 0.0004984186661360019,
"loss": 0.3534,
"step": 17800
},
{
"epoch": 1.0148965203547873,
"grad_norm": 0.10367763787508011,
"learning_rate": 0.0004984142241869457,
"loss": 0.3564,
"step": 17850
},
{
"epoch": 1.0177393677507391,
"grad_norm": 0.11785215884447098,
"learning_rate": 0.0004984097822378895,
"loss": 0.3584,
"step": 17900
},
{
"epoch": 1.020582215146691,
"grad_norm": 0.09687703847885132,
"learning_rate": 0.0004984053402888333,
"loss": 0.356,
"step": 17950
},
{
"epoch": 1.0234250625426426,
"grad_norm": 0.1126837283372879,
"learning_rate": 0.0004984008983397771,
"loss": 0.3545,
"step": 18000
},
{
"epoch": 1.0262679099385945,
"grad_norm": 0.0994856208562851,
"learning_rate": 0.0004983964563907209,
"loss": 0.3556,
"step": 18050
},
{
"epoch": 1.0291107573345464,
"grad_norm": 0.09857575595378876,
"learning_rate": 0.0004983920144416647,
"loss": 0.3596,
"step": 18100
},
{
"epoch": 1.031953604730498,
"grad_norm": 0.10661829262971878,
"learning_rate": 0.0004983875724926086,
"loss": 0.3557,
"step": 18150
},
{
"epoch": 1.0347964521264499,
"grad_norm": 0.08805033564567566,
"learning_rate": 0.0004983831305435525,
"loss": 0.3552,
"step": 18200
},
{
"epoch": 1.0376392995224015,
"grad_norm": 0.09851615130901337,
"learning_rate": 0.0004983786885944963,
"loss": 0.3595,
"step": 18250
},
{
"epoch": 1.0404821469183534,
"grad_norm": 0.10526396334171295,
"learning_rate": 0.0004983742466454401,
"loss": 0.3557,
"step": 18300
},
{
"epoch": 1.0433249943143053,
"grad_norm": 0.10095933824777603,
"learning_rate": 0.0004983698046963839,
"loss": 0.3545,
"step": 18350
},
{
"epoch": 1.046167841710257,
"grad_norm": 0.0906311646103859,
"learning_rate": 0.0004983653627473278,
"loss": 0.3539,
"step": 18400
},
{
"epoch": 1.0490106891062088,
"grad_norm": 0.10459648072719574,
"learning_rate": 0.0004983609207982716,
"loss": 0.3551,
"step": 18450
},
{
"epoch": 1.0518535365021606,
"grad_norm": 0.08786173164844513,
"learning_rate": 0.0004983564788492154,
"loss": 0.3534,
"step": 18500
},
{
"epoch": 1.0546963838981123,
"grad_norm": 0.10580555349588394,
"learning_rate": 0.0004983520369001592,
"loss": 0.3552,
"step": 18550
},
{
"epoch": 1.0575392312940641,
"grad_norm": 0.09845657646656036,
"learning_rate": 0.000498347594951103,
"loss": 0.3567,
"step": 18600
},
{
"epoch": 1.060382078690016,
"grad_norm": 0.09070323407649994,
"learning_rate": 0.0004983431530020468,
"loss": 0.3572,
"step": 18650
},
{
"epoch": 1.0632249260859676,
"grad_norm": 0.11504799127578735,
"learning_rate": 0.0004983387110529907,
"loss": 0.3546,
"step": 18700
},
{
"epoch": 1.0660677734819195,
"grad_norm": 0.0939740464091301,
"learning_rate": 0.0004983342691039346,
"loss": 0.3572,
"step": 18750
},
{
"epoch": 1.0689106208778714,
"grad_norm": 0.11082996428012848,
"learning_rate": 0.0004983298271548784,
"loss": 0.357,
"step": 18800
},
{
"epoch": 1.071753468273823,
"grad_norm": 0.09426229447126389,
"learning_rate": 0.0004983253852058222,
"loss": 0.3552,
"step": 18850
},
{
"epoch": 1.0745963156697749,
"grad_norm": 0.09430352598428726,
"learning_rate": 0.000498320943256766,
"loss": 0.3559,
"step": 18900
},
{
"epoch": 1.0774391630657267,
"grad_norm": 0.11377741396427155,
"learning_rate": 0.0004983165013077098,
"loss": 0.3517,
"step": 18950
},
{
"epoch": 1.0802820104616784,
"grad_norm": 0.0836120992898941,
"learning_rate": 0.0004983120593586536,
"loss": 0.3546,
"step": 19000
},
{
"epoch": 1.0831248578576302,
"grad_norm": 0.09483205527067184,
"learning_rate": 0.0004983076174095974,
"loss": 0.354,
"step": 19050
},
{
"epoch": 1.0859677052535819,
"grad_norm": 0.08619007468223572,
"learning_rate": 0.0004983031754605412,
"loss": 0.3526,
"step": 19100
},
{
"epoch": 1.0888105526495337,
"grad_norm": 0.09521963447332382,
"learning_rate": 0.0004982987335114852,
"loss": 0.3542,
"step": 19150
},
{
"epoch": 1.0916534000454856,
"grad_norm": 0.08765676617622375,
"learning_rate": 0.000498294291562429,
"loss": 0.3542,
"step": 19200
},
{
"epoch": 1.0944962474414373,
"grad_norm": 0.10146308690309525,
"learning_rate": 0.0004982898496133728,
"loss": 0.3559,
"step": 19250
},
{
"epoch": 1.0973390948373891,
"grad_norm": 0.09321428835391998,
"learning_rate": 0.0004982854076643166,
"loss": 0.3576,
"step": 19300
},
{
"epoch": 1.100181942233341,
"grad_norm": 0.11401006579399109,
"learning_rate": 0.0004982809657152604,
"loss": 0.3549,
"step": 19350
},
{
"epoch": 1.1030247896292926,
"grad_norm": 0.11573155969381332,
"learning_rate": 0.0004982765237662042,
"loss": 0.3569,
"step": 19400
},
{
"epoch": 1.1058676370252445,
"grad_norm": 0.08966556191444397,
"learning_rate": 0.000498272081817148,
"loss": 0.3603,
"step": 19450
},
{
"epoch": 1.1087104844211964,
"grad_norm": 0.12934494018554688,
"learning_rate": 0.0004982676398680918,
"loss": 0.3533,
"step": 19500
},
{
"epoch": 1.111553331817148,
"grad_norm": 0.13741086423397064,
"learning_rate": 0.0004982631979190357,
"loss": 0.3542,
"step": 19550
},
{
"epoch": 1.1143961792130999,
"grad_norm": 0.10928856581449509,
"learning_rate": 0.0004982587559699796,
"loss": 0.3559,
"step": 19600
},
{
"epoch": 1.1172390266090517,
"grad_norm": 0.09697018563747406,
"learning_rate": 0.0004982543140209234,
"loss": 0.3559,
"step": 19650
},
{
"epoch": 1.1200818740050034,
"grad_norm": 0.09221459925174713,
"learning_rate": 0.0004982498720718672,
"loss": 0.3554,
"step": 19700
},
{
"epoch": 1.1229247214009552,
"grad_norm": 0.091732919216156,
"learning_rate": 0.0004982454301228111,
"loss": 0.352,
"step": 19750
},
{
"epoch": 1.125767568796907,
"grad_norm": 0.10662009567022324,
"learning_rate": 0.0004982409881737549,
"loss": 0.3582,
"step": 19800
},
{
"epoch": 1.1286104161928587,
"grad_norm": 0.08810966461896896,
"learning_rate": 0.0004982365462246987,
"loss": 0.3534,
"step": 19850
},
{
"epoch": 1.1314532635888106,
"grad_norm": 0.10295290499925613,
"learning_rate": 0.0004982321042756425,
"loss": 0.3577,
"step": 19900
},
{
"epoch": 1.1342961109847622,
"grad_norm": 0.08759327977895737,
"learning_rate": 0.0004982276623265863,
"loss": 0.3529,
"step": 19950
},
{
"epoch": 1.137138958380714,
"grad_norm": 0.08622145652770996,
"learning_rate": 0.0004982232203775301,
"loss": 0.3541,
"step": 20000
},
{
"epoch": 1.137138958380714,
"eval_loss": 0.3468220829963684,
"eval_runtime": 575.5127,
"eval_samples_per_second": 229.204,
"eval_steps_per_second": 28.651,
"step": 20000
},
{
"epoch": 1.139981805776666,
"grad_norm": 0.09593155980110168,
"learning_rate": 0.000498218778428474,
"loss": 0.3533,
"step": 20050
},
{
"epoch": 1.1428246531726176,
"grad_norm": 0.1172272190451622,
"learning_rate": 0.0004982143364794178,
"loss": 0.3513,
"step": 20100
},
{
"epoch": 1.1456675005685695,
"grad_norm": 0.10402148216962814,
"learning_rate": 0.0004982098945303617,
"loss": 0.3587,
"step": 20150
},
{
"epoch": 1.1485103479645213,
"grad_norm": 0.14057371020317078,
"learning_rate": 0.0004982054525813055,
"loss": 0.3525,
"step": 20200
},
{
"epoch": 1.151353195360473,
"grad_norm": 0.11363080888986588,
"learning_rate": 0.0004982010106322493,
"loss": 0.3527,
"step": 20250
},
{
"epoch": 1.1541960427564248,
"grad_norm": 0.10476770997047424,
"learning_rate": 0.0004981965686831931,
"loss": 0.354,
"step": 20300
},
{
"epoch": 1.1570388901523767,
"grad_norm": 0.11272389441728592,
"learning_rate": 0.0004981921267341369,
"loss": 0.3562,
"step": 20350
},
{
"epoch": 1.1598817375483284,
"grad_norm": 0.0867941826581955,
"learning_rate": 0.0004981876847850807,
"loss": 0.3531,
"step": 20400
},
{
"epoch": 1.1627245849442802,
"grad_norm": 0.12125921249389648,
"learning_rate": 0.0004981832428360245,
"loss": 0.3523,
"step": 20450
},
{
"epoch": 1.1655674323402319,
"grad_norm": 0.09518919885158539,
"learning_rate": 0.0004981788008869683,
"loss": 0.3562,
"step": 20500
},
{
"epoch": 1.1684102797361837,
"grad_norm": 0.09472860395908356,
"learning_rate": 0.0004981743589379123,
"loss": 0.3555,
"step": 20550
},
{
"epoch": 1.1712531271321356,
"grad_norm": 0.09812992066144943,
"learning_rate": 0.0004981699169888561,
"loss": 0.3559,
"step": 20600
},
{
"epoch": 1.1740959745280874,
"grad_norm": 0.09982824325561523,
"learning_rate": 0.0004981654750397999,
"loss": 0.3511,
"step": 20650
},
{
"epoch": 1.176938821924039,
"grad_norm": 0.11572203040122986,
"learning_rate": 0.0004981610330907437,
"loss": 0.3524,
"step": 20700
},
{
"epoch": 1.179781669319991,
"grad_norm": 0.113502636551857,
"learning_rate": 0.0004981565911416875,
"loss": 0.3526,
"step": 20750
},
{
"epoch": 1.1826245167159426,
"grad_norm": 0.09475291520357132,
"learning_rate": 0.0004981521491926313,
"loss": 0.3536,
"step": 20800
},
{
"epoch": 1.1854673641118945,
"grad_norm": 0.09980995953083038,
"learning_rate": 0.0004981477072435751,
"loss": 0.3525,
"step": 20850
},
{
"epoch": 1.1883102115078463,
"grad_norm": 0.09389813244342804,
"learning_rate": 0.0004981432652945189,
"loss": 0.3536,
"step": 20900
},
{
"epoch": 1.191153058903798,
"grad_norm": 0.09969864040613174,
"learning_rate": 0.0004981388233454628,
"loss": 0.3536,
"step": 20950
},
{
"epoch": 1.1939959062997498,
"grad_norm": 0.08675362169742584,
"learning_rate": 0.0004981343813964067,
"loss": 0.3538,
"step": 21000
},
{
"epoch": 1.1968387536957017,
"grad_norm": 0.13347360491752625,
"learning_rate": 0.0004981299394473505,
"loss": 0.3536,
"step": 21050
},
{
"epoch": 1.1996816010916533,
"grad_norm": 0.1156553402543068,
"learning_rate": 0.0004981254974982943,
"loss": 0.3516,
"step": 21100
},
{
"epoch": 1.2025244484876052,
"grad_norm": 0.08983340859413147,
"learning_rate": 0.0004981210555492382,
"loss": 0.3537,
"step": 21150
},
{
"epoch": 1.205367295883557,
"grad_norm": 0.0790681466460228,
"learning_rate": 0.000498116613600182,
"loss": 0.3527,
"step": 21200
},
{
"epoch": 1.2082101432795087,
"grad_norm": 0.12216062098741531,
"learning_rate": 0.0004981121716511258,
"loss": 0.3535,
"step": 21250
},
{
"epoch": 1.2110529906754606,
"grad_norm": 0.08232084661722183,
"learning_rate": 0.0004981077297020696,
"loss": 0.3543,
"step": 21300
},
{
"epoch": 1.2138958380714122,
"grad_norm": 0.09808045625686646,
"learning_rate": 0.0004981032877530134,
"loss": 0.3526,
"step": 21350
},
{
"epoch": 1.216738685467364,
"grad_norm": 0.0974017009139061,
"learning_rate": 0.0004980988458039572,
"loss": 0.3518,
"step": 21400
},
{
"epoch": 1.219581532863316,
"grad_norm": 0.11234429478645325,
"learning_rate": 0.000498094403854901,
"loss": 0.3549,
"step": 21450
},
{
"epoch": 1.2224243802592678,
"grad_norm": 0.0913853719830513,
"learning_rate": 0.0004980899619058449,
"loss": 0.3572,
"step": 21500
},
{
"epoch": 1.2252672276552194,
"grad_norm": 0.11108215898275375,
"learning_rate": 0.0004980855199567888,
"loss": 0.3532,
"step": 21550
},
{
"epoch": 1.2281100750511713,
"grad_norm": 0.09476979076862335,
"learning_rate": 0.0004980810780077326,
"loss": 0.3557,
"step": 21600
},
{
"epoch": 1.230952922447123,
"grad_norm": 0.0904000923037529,
"learning_rate": 0.0004980766360586764,
"loss": 0.3512,
"step": 21650
},
{
"epoch": 1.2337957698430748,
"grad_norm": 0.09948475658893585,
"learning_rate": 0.0004980721941096202,
"loss": 0.3533,
"step": 21700
},
{
"epoch": 1.2366386172390267,
"grad_norm": 0.0987270325422287,
"learning_rate": 0.000498067752160564,
"loss": 0.354,
"step": 21750
},
{
"epoch": 1.2394814646349783,
"grad_norm": 0.10672775655984879,
"learning_rate": 0.0004980633102115078,
"loss": 0.3521,
"step": 21800
},
{
"epoch": 1.2423243120309302,
"grad_norm": 0.11016605794429779,
"learning_rate": 0.0004980588682624516,
"loss": 0.3519,
"step": 21850
},
{
"epoch": 1.245167159426882,
"grad_norm": 0.11307205259799957,
"learning_rate": 0.0004980544263133954,
"loss": 0.3545,
"step": 21900
},
{
"epoch": 1.2480100068228337,
"grad_norm": 0.09647481143474579,
"learning_rate": 0.0004980499843643394,
"loss": 0.3497,
"step": 21950
},
{
"epoch": 1.2508528542187856,
"grad_norm": 0.11781589686870575,
"learning_rate": 0.0004980455424152832,
"loss": 0.3527,
"step": 22000
},
{
"epoch": 1.2536957016147374,
"grad_norm": 0.09155919402837753,
"learning_rate": 0.000498041100466227,
"loss": 0.3541,
"step": 22050
},
{
"epoch": 1.256538549010689,
"grad_norm": 0.09770865738391876,
"learning_rate": 0.0004980366585171708,
"loss": 0.351,
"step": 22100
},
{
"epoch": 1.259381396406641,
"grad_norm": 0.09057383239269257,
"learning_rate": 0.0004980322165681146,
"loss": 0.3526,
"step": 22150
},
{
"epoch": 1.2622242438025926,
"grad_norm": 0.10171497613191605,
"learning_rate": 0.0004980277746190584,
"loss": 0.3516,
"step": 22200
},
{
"epoch": 1.2650670911985444,
"grad_norm": 0.12109605222940445,
"learning_rate": 0.0004980233326700022,
"loss": 0.3479,
"step": 22250
},
{
"epoch": 1.2679099385944963,
"grad_norm": 0.08494057506322861,
"learning_rate": 0.000498018890720946,
"loss": 0.3536,
"step": 22300
},
{
"epoch": 1.2707527859904482,
"grad_norm": 0.10715505480766296,
"learning_rate": 0.00049801444877189,
"loss": 0.354,
"step": 22350
},
{
"epoch": 1.2735956333863998,
"grad_norm": 0.09710002690553665,
"learning_rate": 0.0004980100068228338,
"loss": 0.3504,
"step": 22400
},
{
"epoch": 1.2764384807823517,
"grad_norm": 0.10588357597589493,
"learning_rate": 0.0004980055648737776,
"loss": 0.3552,
"step": 22450
},
{
"epoch": 1.2792813281783033,
"grad_norm": 0.09577326476573944,
"learning_rate": 0.0004980011229247214,
"loss": 0.3523,
"step": 22500
},
{
"epoch": 1.2821241755742552,
"grad_norm": 0.0960511788725853,
"learning_rate": 0.0004979966809756653,
"loss": 0.3496,
"step": 22550
},
{
"epoch": 1.284967022970207,
"grad_norm": 0.09112073481082916,
"learning_rate": 0.0004979922390266091,
"loss": 0.3516,
"step": 22600
},
{
"epoch": 1.2878098703661587,
"grad_norm": 0.11254040151834488,
"learning_rate": 0.0004979877970775529,
"loss": 0.3508,
"step": 22650
},
{
"epoch": 1.2906527177621105,
"grad_norm": 0.1305217444896698,
"learning_rate": 0.0004979833551284967,
"loss": 0.3484,
"step": 22700
},
{
"epoch": 1.2934955651580622,
"grad_norm": 0.11914518475532532,
"learning_rate": 0.0004979789131794405,
"loss": 0.3492,
"step": 22750
},
{
"epoch": 1.296338412554014,
"grad_norm": 0.10428331047296524,
"learning_rate": 0.0004979744712303843,
"loss": 0.3516,
"step": 22800
},
{
"epoch": 1.299181259949966,
"grad_norm": 0.10348132997751236,
"learning_rate": 0.0004979700292813282,
"loss": 0.3497,
"step": 22850
},
{
"epoch": 1.3020241073459178,
"grad_norm": 0.08855703473091125,
"learning_rate": 0.0004979655873322721,
"loss": 0.3516,
"step": 22900
},
{
"epoch": 1.3048669547418694,
"grad_norm": 0.11277921497821808,
"learning_rate": 0.0004979611453832159,
"loss": 0.3473,
"step": 22950
},
{
"epoch": 1.3077098021378213,
"grad_norm": 0.092954620718956,
"learning_rate": 0.0004979567034341597,
"loss": 0.3483,
"step": 23000
},
{
"epoch": 1.310552649533773,
"grad_norm": 0.11179167777299881,
"learning_rate": 0.0004979522614851035,
"loss": 0.3512,
"step": 23050
},
{
"epoch": 1.3133954969297248,
"grad_norm": 0.09125496447086334,
"learning_rate": 0.0004979478195360473,
"loss": 0.3508,
"step": 23100
},
{
"epoch": 1.3162383443256767,
"grad_norm": 0.0916043147444725,
"learning_rate": 0.0004979433775869911,
"loss": 0.3523,
"step": 23150
},
{
"epoch": 1.3190811917216285,
"grad_norm": 0.09685220569372177,
"learning_rate": 0.0004979389356379349,
"loss": 0.3546,
"step": 23200
},
{
"epoch": 1.3219240391175802,
"grad_norm": 0.10811689496040344,
"learning_rate": 0.0004979344936888787,
"loss": 0.3498,
"step": 23250
},
{
"epoch": 1.324766886513532,
"grad_norm": 0.08741540461778641,
"learning_rate": 0.0004979300517398227,
"loss": 0.3493,
"step": 23300
},
{
"epoch": 1.3276097339094837,
"grad_norm": 0.08877366036176682,
"learning_rate": 0.0004979256097907665,
"loss": 0.3527,
"step": 23350
},
{
"epoch": 1.3304525813054355,
"grad_norm": 0.1235891655087471,
"learning_rate": 0.0004979211678417103,
"loss": 0.3525,
"step": 23400
},
{
"epoch": 1.3332954287013874,
"grad_norm": 0.0945284515619278,
"learning_rate": 0.0004979167258926541,
"loss": 0.354,
"step": 23450
},
{
"epoch": 1.336138276097339,
"grad_norm": 0.08477913588285446,
"learning_rate": 0.0004979122839435979,
"loss": 0.3515,
"step": 23500
},
{
"epoch": 1.338981123493291,
"grad_norm": 0.09075385332107544,
"learning_rate": 0.0004979078419945417,
"loss": 0.3515,
"step": 23550
},
{
"epoch": 1.3418239708892425,
"grad_norm": 0.08231678605079651,
"learning_rate": 0.0004979034000454855,
"loss": 0.3506,
"step": 23600
},
{
"epoch": 1.3446668182851944,
"grad_norm": 0.08512122184038162,
"learning_rate": 0.0004978989580964293,
"loss": 0.3506,
"step": 23650
},
{
"epoch": 1.3475096656811463,
"grad_norm": 0.09195214509963989,
"learning_rate": 0.0004978945161473732,
"loss": 0.3493,
"step": 23700
},
{
"epoch": 1.3503525130770981,
"grad_norm": 0.09307179600000381,
"learning_rate": 0.000497890074198317,
"loss": 0.3466,
"step": 23750
},
{
"epoch": 1.3531953604730498,
"grad_norm": 0.08473914861679077,
"learning_rate": 0.0004978856322492609,
"loss": 0.3512,
"step": 23800
},
{
"epoch": 1.3560382078690016,
"grad_norm": 0.09676692634820938,
"learning_rate": 0.0004978811903002047,
"loss": 0.3495,
"step": 23850
},
{
"epoch": 1.3588810552649533,
"grad_norm": 0.08961619436740875,
"learning_rate": 0.0004978767483511486,
"loss": 0.3493,
"step": 23900
},
{
"epoch": 1.3617239026609051,
"grad_norm": 0.10256984084844589,
"learning_rate": 0.0004978723064020924,
"loss": 0.3519,
"step": 23950
},
{
"epoch": 1.364566750056857,
"grad_norm": 0.09426955133676529,
"learning_rate": 0.0004978678644530362,
"loss": 0.3496,
"step": 24000
},
{
"epoch": 1.3674095974528087,
"grad_norm": 0.09407513588666916,
"learning_rate": 0.00049786342250398,
"loss": 0.35,
"step": 24050
},
{
"epoch": 1.3702524448487605,
"grad_norm": 0.11077912896871567,
"learning_rate": 0.0004978589805549238,
"loss": 0.3499,
"step": 24100
},
{
"epoch": 1.3730952922447124,
"grad_norm": 0.09664590656757355,
"learning_rate": 0.0004978545386058676,
"loss": 0.3509,
"step": 24150
},
{
"epoch": 1.375938139640664,
"grad_norm": 0.08676467835903168,
"learning_rate": 0.0004978500966568114,
"loss": 0.3498,
"step": 24200
},
{
"epoch": 1.378780987036616,
"grad_norm": 0.07996726781129837,
"learning_rate": 0.0004978456547077553,
"loss": 0.3525,
"step": 24250
},
{
"epoch": 1.3816238344325678,
"grad_norm": 0.1071229949593544,
"learning_rate": 0.0004978412127586992,
"loss": 0.3529,
"step": 24300
},
{
"epoch": 1.3844666818285194,
"grad_norm": 0.09521368145942688,
"learning_rate": 0.000497836770809643,
"loss": 0.3501,
"step": 24350
},
{
"epoch": 1.3873095292244713,
"grad_norm": 0.10842622071504593,
"learning_rate": 0.0004978323288605868,
"loss": 0.3469,
"step": 24400
},
{
"epoch": 1.390152376620423,
"grad_norm": 0.08254586160182953,
"learning_rate": 0.0004978278869115306,
"loss": 0.3481,
"step": 24450
},
{
"epoch": 1.3929952240163748,
"grad_norm": 0.122976154088974,
"learning_rate": 0.0004978234449624744,
"loss": 0.3519,
"step": 24500
},
{
"epoch": 1.3958380714123266,
"grad_norm": 0.10202399641275406,
"learning_rate": 0.0004978190030134182,
"loss": 0.3481,
"step": 24550
},
{
"epoch": 1.3986809188082785,
"grad_norm": 0.09278547018766403,
"learning_rate": 0.000497814561064362,
"loss": 0.3477,
"step": 24600
},
{
"epoch": 1.4015237662042301,
"grad_norm": 0.09204497933387756,
"learning_rate": 0.0004978101191153058,
"loss": 0.3487,
"step": 24650
},
{
"epoch": 1.404366613600182,
"grad_norm": 0.09469152987003326,
"learning_rate": 0.0004978056771662498,
"loss": 0.352,
"step": 24700
},
{
"epoch": 1.4072094609961336,
"grad_norm": 0.09524402767419815,
"learning_rate": 0.0004978012352171936,
"loss": 0.3502,
"step": 24750
},
{
"epoch": 1.4100523083920855,
"grad_norm": 0.09983129799365997,
"learning_rate": 0.0004977967932681374,
"loss": 0.3491,
"step": 24800
},
{
"epoch": 1.4128951557880374,
"grad_norm": 0.13477516174316406,
"learning_rate": 0.0004977923513190812,
"loss": 0.3505,
"step": 24850
},
{
"epoch": 1.415738003183989,
"grad_norm": 0.09205208718776703,
"learning_rate": 0.000497787909370025,
"loss": 0.3505,
"step": 24900
},
{
"epoch": 1.4185808505799409,
"grad_norm": 0.08766143023967743,
"learning_rate": 0.0004977834674209688,
"loss": 0.3508,
"step": 24950
},
{
"epoch": 1.4214236979758925,
"grad_norm": 0.11106764525175095,
"learning_rate": 0.0004977790254719127,
"loss": 0.3501,
"step": 25000
},
{
"epoch": 1.4214236979758925,
"eval_loss": 0.3424564003944397,
"eval_runtime": 577.5935,
"eval_samples_per_second": 228.379,
"eval_steps_per_second": 28.548,
"step": 25000
}
],
"logging_steps": 50,
"max_steps": 5628160,
"num_input_tokens_seen": 0,
"num_train_epochs": 320,
"save_steps": 5000,
"stateful_callbacks": {
"EarlyStoppingCallback": {
"args": {
"early_stopping_patience": 3,
"early_stopping_threshold": 0.0
},
"attributes": {
"early_stopping_patience_counter": 0
}
},
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 9.134362268821094e+17,
"train_batch_size": 120,
"trial_name": null,
"trial_params": null
}