Qwen_2.5-Math-v3 / trainer_state.json
QuangDuy's picture
Add checkpoint-4992
d2ec5d4 verified
raw
history blame
87.8 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"eval_steps": 500,
"global_step": 4992,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.004006410256410256,
"grad_norm": 1.243200421333313,
"learning_rate": 6.666666666666667e-06,
"loss": 0.5025,
"step": 10
},
{
"epoch": 0.008012820512820512,
"grad_norm": 0.40575382113456726,
"learning_rate": 1.3333333333333333e-05,
"loss": 0.3957,
"step": 20
},
{
"epoch": 0.01201923076923077,
"grad_norm": 0.2619800269603729,
"learning_rate": 2e-05,
"loss": 0.3277,
"step": 30
},
{
"epoch": 0.016025641025641024,
"grad_norm": 0.2629871964454651,
"learning_rate": 2.6666666666666667e-05,
"loss": 0.319,
"step": 40
},
{
"epoch": 0.020032051282051284,
"grad_norm": 0.2376326471567154,
"learning_rate": 3.3333333333333335e-05,
"loss": 0.2829,
"step": 50
},
{
"epoch": 0.02403846153846154,
"grad_norm": 0.25014540553092957,
"learning_rate": 4e-05,
"loss": 0.2777,
"step": 60
},
{
"epoch": 0.028044871794871796,
"grad_norm": 0.30097439885139465,
"learning_rate": 4.666666666666667e-05,
"loss": 0.2688,
"step": 70
},
{
"epoch": 0.03205128205128205,
"grad_norm": 0.29906877875328064,
"learning_rate": 5.333333333333333e-05,
"loss": 0.272,
"step": 80
},
{
"epoch": 0.036057692307692304,
"grad_norm": 0.31522494554519653,
"learning_rate": 6e-05,
"loss": 0.2573,
"step": 90
},
{
"epoch": 0.04006410256410257,
"grad_norm": 0.31278184056282043,
"learning_rate": 6.666666666666667e-05,
"loss": 0.2548,
"step": 100
},
{
"epoch": 0.044070512820512824,
"grad_norm": 0.31608548760414124,
"learning_rate": 7.333333333333333e-05,
"loss": 0.2496,
"step": 110
},
{
"epoch": 0.04807692307692308,
"grad_norm": 0.29328644275665283,
"learning_rate": 8e-05,
"loss": 0.2475,
"step": 120
},
{
"epoch": 0.052083333333333336,
"grad_norm": 0.33972907066345215,
"learning_rate": 8.666666666666667e-05,
"loss": 0.2512,
"step": 130
},
{
"epoch": 0.05608974358974359,
"grad_norm": 0.32710641622543335,
"learning_rate": 9.333333333333334e-05,
"loss": 0.2382,
"step": 140
},
{
"epoch": 0.06009615384615385,
"grad_norm": 0.2911882996559143,
"learning_rate": 0.0001,
"loss": 0.2373,
"step": 150
},
{
"epoch": 0.0641025641025641,
"grad_norm": 0.29396378993988037,
"learning_rate": 9.999894758104792e-05,
"loss": 0.2433,
"step": 160
},
{
"epoch": 0.06810897435897435,
"grad_norm": 0.33355265855789185,
"learning_rate": 9.999579036849509e-05,
"loss": 0.2435,
"step": 170
},
{
"epoch": 0.07211538461538461,
"grad_norm": 0.29101336002349854,
"learning_rate": 9.999052849524992e-05,
"loss": 0.2376,
"step": 180
},
{
"epoch": 0.07612179487179487,
"grad_norm": 0.27655455470085144,
"learning_rate": 9.998316218282024e-05,
"loss": 0.2365,
"step": 190
},
{
"epoch": 0.08012820512820513,
"grad_norm": 0.2681509554386139,
"learning_rate": 9.997369174130391e-05,
"loss": 0.2425,
"step": 200
},
{
"epoch": 0.08413461538461539,
"grad_norm": 0.26265430450439453,
"learning_rate": 9.996211756937579e-05,
"loss": 0.2389,
"step": 210
},
{
"epoch": 0.08814102564102565,
"grad_norm": 0.2830798625946045,
"learning_rate": 9.994844015427102e-05,
"loss": 0.2304,
"step": 220
},
{
"epoch": 0.0921474358974359,
"grad_norm": 0.2548621892929077,
"learning_rate": 9.993266007176445e-05,
"loss": 0.2308,
"step": 230
},
{
"epoch": 0.09615384615384616,
"grad_norm": 0.27960866689682007,
"learning_rate": 9.991477798614638e-05,
"loss": 0.2277,
"step": 240
},
{
"epoch": 0.10016025641025642,
"grad_norm": 0.2278411090373993,
"learning_rate": 9.989479465019462e-05,
"loss": 0.2314,
"step": 250
},
{
"epoch": 0.10416666666666667,
"grad_norm": 0.2582249045372009,
"learning_rate": 9.987271090514288e-05,
"loss": 0.2355,
"step": 260
},
{
"epoch": 0.10817307692307693,
"grad_norm": 0.28811654448509216,
"learning_rate": 9.984852768064516e-05,
"loss": 0.2285,
"step": 270
},
{
"epoch": 0.11217948717948718,
"grad_norm": 0.241895392537117,
"learning_rate": 9.982224599473689e-05,
"loss": 0.2344,
"step": 280
},
{
"epoch": 0.11618589743589744,
"grad_norm": 0.2796310484409332,
"learning_rate": 9.97938669537918e-05,
"loss": 0.2246,
"step": 290
},
{
"epoch": 0.1201923076923077,
"grad_norm": 0.24658632278442383,
"learning_rate": 9.976339175247551e-05,
"loss": 0.2187,
"step": 300
},
{
"epoch": 0.12419871794871795,
"grad_norm": 0.23214490711688995,
"learning_rate": 9.973082167369521e-05,
"loss": 0.225,
"step": 310
},
{
"epoch": 0.1282051282051282,
"grad_norm": 0.25273966789245605,
"learning_rate": 9.969615808854562e-05,
"loss": 0.2291,
"step": 320
},
{
"epoch": 0.13221153846153846,
"grad_norm": 0.25187525153160095,
"learning_rate": 9.965940245625131e-05,
"loss": 0.2227,
"step": 330
},
{
"epoch": 0.1362179487179487,
"grad_norm": 0.296224445104599,
"learning_rate": 9.962055632410523e-05,
"loss": 0.2185,
"step": 340
},
{
"epoch": 0.14022435897435898,
"grad_norm": 0.2560974955558777,
"learning_rate": 9.95796213274036e-05,
"loss": 0.2251,
"step": 350
},
{
"epoch": 0.14423076923076922,
"grad_norm": 0.24337032437324524,
"learning_rate": 9.95365991893771e-05,
"loss": 0.2212,
"step": 360
},
{
"epoch": 0.1482371794871795,
"grad_norm": 0.2607037127017975,
"learning_rate": 9.949149172111825e-05,
"loss": 0.2229,
"step": 370
},
{
"epoch": 0.15224358974358973,
"grad_norm": 0.2829968333244324,
"learning_rate": 9.944430082150522e-05,
"loss": 0.2276,
"step": 380
},
{
"epoch": 0.15625,
"grad_norm": 0.24074934422969818,
"learning_rate": 9.93950284771219e-05,
"loss": 0.2241,
"step": 390
},
{
"epoch": 0.16025641025641027,
"grad_norm": 0.27647778391838074,
"learning_rate": 9.934367676217427e-05,
"loss": 0.2071,
"step": 400
},
{
"epoch": 0.1642628205128205,
"grad_norm": 0.2719784677028656,
"learning_rate": 9.929024783840302e-05,
"loss": 0.2217,
"step": 410
},
{
"epoch": 0.16826923076923078,
"grad_norm": 0.23359382152557373,
"learning_rate": 9.923474395499265e-05,
"loss": 0.2222,
"step": 420
},
{
"epoch": 0.17227564102564102,
"grad_norm": 0.27376872301101685,
"learning_rate": 9.917716744847673e-05,
"loss": 0.222,
"step": 430
},
{
"epoch": 0.1762820512820513,
"grad_norm": 0.2602153718471527,
"learning_rate": 9.911752074263949e-05,
"loss": 0.2232,
"step": 440
},
{
"epoch": 0.18028846153846154,
"grad_norm": 0.29065001010894775,
"learning_rate": 9.905580634841388e-05,
"loss": 0.2212,
"step": 450
},
{
"epoch": 0.1842948717948718,
"grad_norm": 0.24715092778205872,
"learning_rate": 9.899202686377584e-05,
"loss": 0.2246,
"step": 460
},
{
"epoch": 0.18830128205128205,
"grad_norm": 0.2442002296447754,
"learning_rate": 9.89261849736349e-05,
"loss": 0.2171,
"step": 470
},
{
"epoch": 0.19230769230769232,
"grad_norm": 0.22722263634204865,
"learning_rate": 9.88582834497212e-05,
"loss": 0.2176,
"step": 480
},
{
"epoch": 0.19631410256410256,
"grad_norm": 0.22697150707244873,
"learning_rate": 9.878832515046871e-05,
"loss": 0.2179,
"step": 490
},
{
"epoch": 0.20032051282051283,
"grad_norm": 0.23478808999061584,
"learning_rate": 9.87163130208951e-05,
"loss": 0.2114,
"step": 500
},
{
"epoch": 0.20432692307692307,
"grad_norm": 0.26591965556144714,
"learning_rate": 9.864225009247751e-05,
"loss": 0.2165,
"step": 510
},
{
"epoch": 0.20833333333333334,
"grad_norm": 0.23949167132377625,
"learning_rate": 9.856613948302518e-05,
"loss": 0.2118,
"step": 520
},
{
"epoch": 0.21233974358974358,
"grad_norm": 0.22695676982402802,
"learning_rate": 9.848798439654797e-05,
"loss": 0.2256,
"step": 530
},
{
"epoch": 0.21634615384615385,
"grad_norm": 0.2355981171131134,
"learning_rate": 9.840778812312165e-05,
"loss": 0.2123,
"step": 540
},
{
"epoch": 0.2203525641025641,
"grad_norm": 0.2660483419895172,
"learning_rate": 9.832555403874937e-05,
"loss": 0.2207,
"step": 550
},
{
"epoch": 0.22435897435897437,
"grad_norm": 0.2517046630382538,
"learning_rate": 9.824128560521948e-05,
"loss": 0.2144,
"step": 560
},
{
"epoch": 0.2283653846153846,
"grad_norm": 0.23771142959594727,
"learning_rate": 9.815498636995983e-05,
"loss": 0.2061,
"step": 570
},
{
"epoch": 0.23237179487179488,
"grad_norm": 0.23048646748065948,
"learning_rate": 9.806665996588844e-05,
"loss": 0.2151,
"step": 580
},
{
"epoch": 0.23637820512820512,
"grad_norm": 0.26294463872909546,
"learning_rate": 9.79763101112606e-05,
"loss": 0.2129,
"step": 590
},
{
"epoch": 0.2403846153846154,
"grad_norm": 0.2391149401664734,
"learning_rate": 9.788394060951229e-05,
"loss": 0.2134,
"step": 600
},
{
"epoch": 0.24439102564102563,
"grad_norm": 0.21405121684074402,
"learning_rate": 9.778955534910003e-05,
"loss": 0.2204,
"step": 610
},
{
"epoch": 0.2483974358974359,
"grad_norm": 0.21098735928535461,
"learning_rate": 9.769315830333735e-05,
"loss": 0.212,
"step": 620
},
{
"epoch": 0.25240384615384615,
"grad_norm": 0.24548964202404022,
"learning_rate": 9.759475353022734e-05,
"loss": 0.2091,
"step": 630
},
{
"epoch": 0.2564102564102564,
"grad_norm": 0.23211954534053802,
"learning_rate": 9.749434517229191e-05,
"loss": 0.2074,
"step": 640
},
{
"epoch": 0.2604166666666667,
"grad_norm": 0.24367853999137878,
"learning_rate": 9.739193745639745e-05,
"loss": 0.2,
"step": 650
},
{
"epoch": 0.2644230769230769,
"grad_norm": 0.22633159160614014,
"learning_rate": 9.728753469357677e-05,
"loss": 0.2144,
"step": 660
},
{
"epoch": 0.26842948717948717,
"grad_norm": 0.23814593255519867,
"learning_rate": 9.718114127884774e-05,
"loss": 0.2105,
"step": 670
},
{
"epoch": 0.2724358974358974,
"grad_norm": 0.23060369491577148,
"learning_rate": 9.70727616910282e-05,
"loss": 0.2095,
"step": 680
},
{
"epoch": 0.2764423076923077,
"grad_norm": 0.24806129932403564,
"learning_rate": 9.696240049254743e-05,
"loss": 0.2023,
"step": 690
},
{
"epoch": 0.28044871794871795,
"grad_norm": 0.2317938357591629,
"learning_rate": 9.685006232925413e-05,
"loss": 0.2069,
"step": 700
},
{
"epoch": 0.2844551282051282,
"grad_norm": 0.2503266930580139,
"learning_rate": 9.673575193022073e-05,
"loss": 0.2131,
"step": 710
},
{
"epoch": 0.28846153846153844,
"grad_norm": 0.24147148430347443,
"learning_rate": 9.661947410754449e-05,
"loss": 0.2024,
"step": 720
},
{
"epoch": 0.29246794871794873,
"grad_norm": 0.2683294117450714,
"learning_rate": 9.650123375614477e-05,
"loss": 0.2088,
"step": 730
},
{
"epoch": 0.296474358974359,
"grad_norm": 0.23034317791461945,
"learning_rate": 9.638103585355705e-05,
"loss": 0.2103,
"step": 740
},
{
"epoch": 0.3004807692307692,
"grad_norm": 0.23943544924259186,
"learning_rate": 9.625888545972332e-05,
"loss": 0.211,
"step": 750
},
{
"epoch": 0.30448717948717946,
"grad_norm": 0.24953238666057587,
"learning_rate": 9.613478771677921e-05,
"loss": 0.217,
"step": 760
},
{
"epoch": 0.30849358974358976,
"grad_norm": 0.25620102882385254,
"learning_rate": 9.600874784883736e-05,
"loss": 0.208,
"step": 770
},
{
"epoch": 0.3125,
"grad_norm": 0.22348496317863464,
"learning_rate": 9.588077116176756e-05,
"loss": 0.2048,
"step": 780
},
{
"epoch": 0.31650641025641024,
"grad_norm": 0.21532465517520905,
"learning_rate": 9.575086304297351e-05,
"loss": 0.1997,
"step": 790
},
{
"epoch": 0.32051282051282054,
"grad_norm": 0.23346661031246185,
"learning_rate": 9.561902896116583e-05,
"loss": 0.2085,
"step": 800
},
{
"epoch": 0.3245192307692308,
"grad_norm": 0.23158910870552063,
"learning_rate": 9.548527446613198e-05,
"loss": 0.2005,
"step": 810
},
{
"epoch": 0.328525641025641,
"grad_norm": 0.2302473783493042,
"learning_rate": 9.534960518850257e-05,
"loss": 0.1993,
"step": 820
},
{
"epoch": 0.33253205128205127,
"grad_norm": 0.21067394316196442,
"learning_rate": 9.521202683951436e-05,
"loss": 0.1902,
"step": 830
},
{
"epoch": 0.33653846153846156,
"grad_norm": 0.22453920543193817,
"learning_rate": 9.507254521076982e-05,
"loss": 0.1907,
"step": 840
},
{
"epoch": 0.3405448717948718,
"grad_norm": 0.23678337037563324,
"learning_rate": 9.493116617399334e-05,
"loss": 0.2007,
"step": 850
},
{
"epoch": 0.34455128205128205,
"grad_norm": 0.2377959042787552,
"learning_rate": 9.478789568078403e-05,
"loss": 0.2077,
"step": 860
},
{
"epoch": 0.3485576923076923,
"grad_norm": 0.23520340025424957,
"learning_rate": 9.464273976236517e-05,
"loss": 0.2079,
"step": 870
},
{
"epoch": 0.3525641025641026,
"grad_norm": 0.21978124976158142,
"learning_rate": 9.449570452933038e-05,
"loss": 0.2076,
"step": 880
},
{
"epoch": 0.35657051282051283,
"grad_norm": 0.21525253355503082,
"learning_rate": 9.434679617138625e-05,
"loss": 0.207,
"step": 890
},
{
"epoch": 0.3605769230769231,
"grad_norm": 0.23039253056049347,
"learning_rate": 9.419602095709193e-05,
"loss": 0.1985,
"step": 900
},
{
"epoch": 0.3645833333333333,
"grad_norm": 0.2329927235841751,
"learning_rate": 9.40433852335951e-05,
"loss": 0.2136,
"step": 910
},
{
"epoch": 0.3685897435897436,
"grad_norm": 0.24603788554668427,
"learning_rate": 9.388889542636495e-05,
"loss": 0.2106,
"step": 920
},
{
"epoch": 0.37259615384615385,
"grad_norm": 0.2563282549381256,
"learning_rate": 9.373255803892149e-05,
"loss": 0.2052,
"step": 930
},
{
"epoch": 0.3766025641025641,
"grad_norm": 0.2391922026872635,
"learning_rate": 9.357437965256188e-05,
"loss": 0.1965,
"step": 940
},
{
"epoch": 0.38060897435897434,
"grad_norm": 0.22608546912670135,
"learning_rate": 9.341436692608341e-05,
"loss": 0.1995,
"step": 950
},
{
"epoch": 0.38461538461538464,
"grad_norm": 0.2254156768321991,
"learning_rate": 9.325252659550309e-05,
"loss": 0.2023,
"step": 960
},
{
"epoch": 0.3886217948717949,
"grad_norm": 0.2172781527042389,
"learning_rate": 9.30888654737742e-05,
"loss": 0.1935,
"step": 970
},
{
"epoch": 0.3926282051282051,
"grad_norm": 0.24400238692760468,
"learning_rate": 9.292339045049938e-05,
"loss": 0.2046,
"step": 980
},
{
"epoch": 0.39663461538461536,
"grad_norm": 0.23503102362155914,
"learning_rate": 9.275610849164064e-05,
"loss": 0.1896,
"step": 990
},
{
"epoch": 0.40064102564102566,
"grad_norm": 0.23404181003570557,
"learning_rate": 9.258702663922614e-05,
"loss": 0.2034,
"step": 1000
},
{
"epoch": 0.4046474358974359,
"grad_norm": 0.24212764203548431,
"learning_rate": 9.241615201105372e-05,
"loss": 0.2027,
"step": 1010
},
{
"epoch": 0.40865384615384615,
"grad_norm": 0.2275129109621048,
"learning_rate": 9.224349180039124e-05,
"loss": 0.195,
"step": 1020
},
{
"epoch": 0.4126602564102564,
"grad_norm": 0.2557786703109741,
"learning_rate": 9.206905327567386e-05,
"loss": 0.2029,
"step": 1030
},
{
"epoch": 0.4166666666666667,
"grad_norm": 0.23480971157550812,
"learning_rate": 9.189284378019794e-05,
"loss": 0.1896,
"step": 1040
},
{
"epoch": 0.4206730769230769,
"grad_norm": 0.20550891757011414,
"learning_rate": 9.171487073181198e-05,
"loss": 0.2012,
"step": 1050
},
{
"epoch": 0.42467948717948717,
"grad_norm": 0.23816007375717163,
"learning_rate": 9.153514162260432e-05,
"loss": 0.2038,
"step": 1060
},
{
"epoch": 0.4286858974358974,
"grad_norm": 0.22253037989139557,
"learning_rate": 9.135366401858783e-05,
"loss": 0.2038,
"step": 1070
},
{
"epoch": 0.4326923076923077,
"grad_norm": 0.2550757825374603,
"learning_rate": 9.117044555938128e-05,
"loss": 0.192,
"step": 1080
},
{
"epoch": 0.43669871794871795,
"grad_norm": 0.24329575896263123,
"learning_rate": 9.098549395788784e-05,
"loss": 0.2033,
"step": 1090
},
{
"epoch": 0.4407051282051282,
"grad_norm": 0.23914198577404022,
"learning_rate": 9.079881699997033e-05,
"loss": 0.203,
"step": 1100
},
{
"epoch": 0.44471153846153844,
"grad_norm": 0.22172652184963226,
"learning_rate": 9.061042254412346e-05,
"loss": 0.1974,
"step": 1110
},
{
"epoch": 0.44871794871794873,
"grad_norm": 0.24050387740135193,
"learning_rate": 9.042031852114311e-05,
"loss": 0.2068,
"step": 1120
},
{
"epoch": 0.452724358974359,
"grad_norm": 0.25631067156791687,
"learning_rate": 9.022851293379231e-05,
"loss": 0.1932,
"step": 1130
},
{
"epoch": 0.4567307692307692,
"grad_norm": 0.24415907263755798,
"learning_rate": 9.003501385646449e-05,
"loss": 0.195,
"step": 1140
},
{
"epoch": 0.46073717948717946,
"grad_norm": 0.2140500694513321,
"learning_rate": 8.983982943484349e-05,
"loss": 0.197,
"step": 1150
},
{
"epoch": 0.46474358974358976,
"grad_norm": 0.2267265021800995,
"learning_rate": 8.964296788556067e-05,
"loss": 0.1956,
"step": 1160
},
{
"epoch": 0.46875,
"grad_norm": 0.2181784212589264,
"learning_rate": 8.944443749584907e-05,
"loss": 0.2025,
"step": 1170
},
{
"epoch": 0.47275641025641024,
"grad_norm": 0.22020593285560608,
"learning_rate": 8.924424662319448e-05,
"loss": 0.1965,
"step": 1180
},
{
"epoch": 0.47676282051282054,
"grad_norm": 0.21737538278102875,
"learning_rate": 8.904240369498362e-05,
"loss": 0.1989,
"step": 1190
},
{
"epoch": 0.4807692307692308,
"grad_norm": 0.21464964747428894,
"learning_rate": 8.883891720814942e-05,
"loss": 0.1845,
"step": 1200
},
{
"epoch": 0.484775641025641,
"grad_norm": 0.26550692319869995,
"learning_rate": 8.863379572881329e-05,
"loss": 0.1888,
"step": 1210
},
{
"epoch": 0.48878205128205127,
"grad_norm": 0.24705030024051666,
"learning_rate": 8.84270478919245e-05,
"loss": 0.1912,
"step": 1220
},
{
"epoch": 0.49278846153846156,
"grad_norm": 0.22611679136753082,
"learning_rate": 8.821868240089676e-05,
"loss": 0.1998,
"step": 1230
},
{
"epoch": 0.4967948717948718,
"grad_norm": 0.21941202878952026,
"learning_rate": 8.800870802724175e-05,
"loss": 0.1981,
"step": 1240
},
{
"epoch": 0.500801282051282,
"grad_norm": 0.22026574611663818,
"learning_rate": 8.779713361019982e-05,
"loss": 0.1915,
"step": 1250
},
{
"epoch": 0.5048076923076923,
"grad_norm": 0.2143823504447937,
"learning_rate": 8.758396805636806e-05,
"loss": 0.1952,
"step": 1260
},
{
"epoch": 0.5088141025641025,
"grad_norm": 0.22727568447589874,
"learning_rate": 8.736922033932522e-05,
"loss": 0.1942,
"step": 1270
},
{
"epoch": 0.5128205128205128,
"grad_norm": 0.226875901222229,
"learning_rate": 8.715289949925401e-05,
"loss": 0.1945,
"step": 1280
},
{
"epoch": 0.5168269230769231,
"grad_norm": 0.1967984139919281,
"learning_rate": 8.693501464256046e-05,
"loss": 0.1888,
"step": 1290
},
{
"epoch": 0.5208333333333334,
"grad_norm": 0.2363027185201645,
"learning_rate": 8.67155749414907e-05,
"loss": 0.1925,
"step": 1300
},
{
"epoch": 0.5248397435897436,
"grad_norm": 0.23106949031352997,
"learning_rate": 8.649458963374473e-05,
"loss": 0.1927,
"step": 1310
},
{
"epoch": 0.5288461538461539,
"grad_norm": 0.20555250346660614,
"learning_rate": 8.62720680220876e-05,
"loss": 0.2296,
"step": 1320
},
{
"epoch": 0.5328525641025641,
"grad_norm": 0.21878638863563538,
"learning_rate": 8.604801947395777e-05,
"loss": 0.1993,
"step": 1330
},
{
"epoch": 0.5368589743589743,
"grad_norm": 0.22356002032756805,
"learning_rate": 8.582245342107273e-05,
"loss": 0.1911,
"step": 1340
},
{
"epoch": 0.5408653846153846,
"grad_norm": 0.23506873846054077,
"learning_rate": 8.55953793590321e-05,
"loss": 0.1879,
"step": 1350
},
{
"epoch": 0.5448717948717948,
"grad_norm": 0.23467378318309784,
"learning_rate": 8.53668068469177e-05,
"loss": 0.1949,
"step": 1360
},
{
"epoch": 0.5488782051282052,
"grad_norm": 0.24210813641548157,
"learning_rate": 8.513674550689128e-05,
"loss": 0.1842,
"step": 1370
},
{
"epoch": 0.5528846153846154,
"grad_norm": 0.2602980434894562,
"learning_rate": 8.49052050237894e-05,
"loss": 0.192,
"step": 1380
},
{
"epoch": 0.5568910256410257,
"grad_norm": 0.23115062713623047,
"learning_rate": 8.467219514471581e-05,
"loss": 0.2008,
"step": 1390
},
{
"epoch": 0.5608974358974359,
"grad_norm": 0.23633736371994019,
"learning_rate": 8.443772567863097e-05,
"loss": 0.1893,
"step": 1400
},
{
"epoch": 0.5649038461538461,
"grad_norm": 0.21265539526939392,
"learning_rate": 8.420180649593929e-05,
"loss": 0.2016,
"step": 1410
},
{
"epoch": 0.5689102564102564,
"grad_norm": 0.21362069249153137,
"learning_rate": 8.396444752807357e-05,
"loss": 0.1861,
"step": 1420
},
{
"epoch": 0.5729166666666666,
"grad_norm": 0.2480146735906601,
"learning_rate": 8.37256587670768e-05,
"loss": 0.1948,
"step": 1430
},
{
"epoch": 0.5769230769230769,
"grad_norm": 0.24672983586788177,
"learning_rate": 8.348545026518171e-05,
"loss": 0.193,
"step": 1440
},
{
"epoch": 0.5809294871794872,
"grad_norm": 0.23038969933986664,
"learning_rate": 8.32438321343875e-05,
"loss": 0.1865,
"step": 1450
},
{
"epoch": 0.5849358974358975,
"grad_norm": 0.23314690589904785,
"learning_rate": 8.300081454603415e-05,
"loss": 0.1863,
"step": 1460
},
{
"epoch": 0.5889423076923077,
"grad_norm": 0.2449977695941925,
"learning_rate": 8.27564077303743e-05,
"loss": 0.199,
"step": 1470
},
{
"epoch": 0.592948717948718,
"grad_norm": 0.22973774373531342,
"learning_rate": 8.251062197614257e-05,
"loss": 0.1818,
"step": 1480
},
{
"epoch": 0.5969551282051282,
"grad_norm": 0.2545977830886841,
"learning_rate": 8.226346763012234e-05,
"loss": 0.1963,
"step": 1490
},
{
"epoch": 0.6009615384615384,
"grad_norm": 0.23034323751926422,
"learning_rate": 8.201495509671037e-05,
"loss": 0.1791,
"step": 1500
},
{
"epoch": 0.6049679487179487,
"grad_norm": 0.22426968812942505,
"learning_rate": 8.176509483747862e-05,
"loss": 0.1941,
"step": 1510
},
{
"epoch": 0.6089743589743589,
"grad_norm": 0.2258072942495346,
"learning_rate": 8.151389737073403e-05,
"loss": 0.195,
"step": 1520
},
{
"epoch": 0.6129807692307693,
"grad_norm": 0.22930273413658142,
"learning_rate": 8.126137327107554e-05,
"loss": 0.191,
"step": 1530
},
{
"epoch": 0.6169871794871795,
"grad_norm": 0.239192396402359,
"learning_rate": 8.100753316894913e-05,
"loss": 0.2073,
"step": 1540
},
{
"epoch": 0.6209935897435898,
"grad_norm": 0.21774733066558838,
"learning_rate": 8.075238775020011e-05,
"loss": 0.1875,
"step": 1550
},
{
"epoch": 0.625,
"grad_norm": 0.21093542873859406,
"learning_rate": 8.04959477556235e-05,
"loss": 0.1942,
"step": 1560
},
{
"epoch": 0.6290064102564102,
"grad_norm": 0.21208257973194122,
"learning_rate": 8.023822398051168e-05,
"loss": 0.1928,
"step": 1570
},
{
"epoch": 0.6330128205128205,
"grad_norm": 0.20604068040847778,
"learning_rate": 7.997922727420011e-05,
"loss": 0.1891,
"step": 1580
},
{
"epoch": 0.6370192307692307,
"grad_norm": 0.209623783826828,
"learning_rate": 7.971896853961042e-05,
"loss": 0.1962,
"step": 1590
},
{
"epoch": 0.6410256410256411,
"grad_norm": 0.22957240045070648,
"learning_rate": 7.945745873279164e-05,
"loss": 0.1908,
"step": 1600
},
{
"epoch": 0.6450320512820513,
"grad_norm": 0.2278134673833847,
"learning_rate": 7.919470886245886e-05,
"loss": 0.186,
"step": 1610
},
{
"epoch": 0.6490384615384616,
"grad_norm": 0.2224593311548233,
"learning_rate": 7.893072998952975e-05,
"loss": 0.1837,
"step": 1620
},
{
"epoch": 0.6530448717948718,
"grad_norm": 0.2841613292694092,
"learning_rate": 7.866553322665911e-05,
"loss": 0.1961,
"step": 1630
},
{
"epoch": 0.657051282051282,
"grad_norm": 0.2239413857460022,
"learning_rate": 7.839912973777088e-05,
"loss": 0.1864,
"step": 1640
},
{
"epoch": 0.6610576923076923,
"grad_norm": 0.24558769166469574,
"learning_rate": 7.813153073758832e-05,
"loss": 0.1902,
"step": 1650
},
{
"epoch": 0.6650641025641025,
"grad_norm": 0.22548821568489075,
"learning_rate": 7.786274749116178e-05,
"loss": 0.1927,
"step": 1660
},
{
"epoch": 0.6690705128205128,
"grad_norm": 0.24735094606876373,
"learning_rate": 7.759279131339454e-05,
"loss": 0.1822,
"step": 1670
},
{
"epoch": 0.6730769230769231,
"grad_norm": 0.23317532241344452,
"learning_rate": 7.732167356856655e-05,
"loss": 0.1928,
"step": 1680
},
{
"epoch": 0.6770833333333334,
"grad_norm": 0.23926283419132233,
"learning_rate": 7.704940566985592e-05,
"loss": 0.1875,
"step": 1690
},
{
"epoch": 0.6810897435897436,
"grad_norm": 0.18621069192886353,
"learning_rate": 7.677599907885851e-05,
"loss": 0.1967,
"step": 1700
},
{
"epoch": 0.6850961538461539,
"grad_norm": 0.2330184280872345,
"learning_rate": 7.650146530510542e-05,
"loss": 0.188,
"step": 1710
},
{
"epoch": 0.6891025641025641,
"grad_norm": 0.2638700008392334,
"learning_rate": 7.622581590557853e-05,
"loss": 0.1815,
"step": 1720
},
{
"epoch": 0.6931089743589743,
"grad_norm": 0.22845664620399475,
"learning_rate": 7.594906248422392e-05,
"loss": 0.1838,
"step": 1730
},
{
"epoch": 0.6971153846153846,
"grad_norm": 0.20396199822425842,
"learning_rate": 7.567121669146343e-05,
"loss": 0.181,
"step": 1740
},
{
"epoch": 0.7011217948717948,
"grad_norm": 0.24452732503414154,
"learning_rate": 7.539229022370417e-05,
"loss": 0.1869,
"step": 1750
},
{
"epoch": 0.7051282051282052,
"grad_norm": 0.21035702526569366,
"learning_rate": 7.511229482284619e-05,
"loss": 0.1912,
"step": 1760
},
{
"epoch": 0.7091346153846154,
"grad_norm": 0.25418221950531006,
"learning_rate": 7.483124227578811e-05,
"loss": 0.1842,
"step": 1770
},
{
"epoch": 0.7131410256410257,
"grad_norm": 0.20252016186714172,
"learning_rate": 7.454914441393104e-05,
"loss": 0.1799,
"step": 1780
},
{
"epoch": 0.7171474358974359,
"grad_norm": 0.22078120708465576,
"learning_rate": 7.426601311268043e-05,
"loss": 0.1864,
"step": 1790
},
{
"epoch": 0.7211538461538461,
"grad_norm": 0.22004160284996033,
"learning_rate": 7.398186029094617e-05,
"loss": 0.1868,
"step": 1800
},
{
"epoch": 0.7251602564102564,
"grad_norm": 0.23584477603435516,
"learning_rate": 7.369669791064084e-05,
"loss": 0.1972,
"step": 1810
},
{
"epoch": 0.7291666666666666,
"grad_norm": 0.23261332511901855,
"learning_rate": 7.341053797617621e-05,
"loss": 0.188,
"step": 1820
},
{
"epoch": 0.7331730769230769,
"grad_norm": 0.23437419533729553,
"learning_rate": 7.312339253395778e-05,
"loss": 0.1912,
"step": 1830
},
{
"epoch": 0.7371794871794872,
"grad_norm": 0.22874341905117035,
"learning_rate": 7.283527367187778e-05,
"loss": 0.1826,
"step": 1840
},
{
"epoch": 0.7411858974358975,
"grad_norm": 0.21897387504577637,
"learning_rate": 7.254619351880625e-05,
"loss": 0.1765,
"step": 1850
},
{
"epoch": 0.7451923076923077,
"grad_norm": 0.24442484974861145,
"learning_rate": 7.225616424408045e-05,
"loss": 0.1903,
"step": 1860
},
{
"epoch": 0.749198717948718,
"grad_norm": 0.22574163973331451,
"learning_rate": 7.19651980569926e-05,
"loss": 0.1861,
"step": 1870
},
{
"epoch": 0.7532051282051282,
"grad_norm": 0.2647426724433899,
"learning_rate": 7.167330720627591e-05,
"loss": 0.1838,
"step": 1880
},
{
"epoch": 0.7572115384615384,
"grad_norm": 0.19669461250305176,
"learning_rate": 7.138050397958888e-05,
"loss": 0.178,
"step": 1890
},
{
"epoch": 0.7612179487179487,
"grad_norm": 0.22219333052635193,
"learning_rate": 7.108680070299813e-05,
"loss": 0.1813,
"step": 1900
},
{
"epoch": 0.7652243589743589,
"grad_norm": 0.24694056808948517,
"learning_rate": 7.079220974045941e-05,
"loss": 0.186,
"step": 1910
},
{
"epoch": 0.7692307692307693,
"grad_norm": 0.2198621928691864,
"learning_rate": 7.049674349329724e-05,
"loss": 0.1883,
"step": 1920
},
{
"epoch": 0.7732371794871795,
"grad_norm": 0.22908103466033936,
"learning_rate": 7.020041439968274e-05,
"loss": 0.1925,
"step": 1930
},
{
"epoch": 0.7772435897435898,
"grad_norm": 0.2108490765094757,
"learning_rate": 6.990323493411006e-05,
"loss": 0.1832,
"step": 1940
},
{
"epoch": 0.78125,
"grad_norm": 0.23686829209327698,
"learning_rate": 6.960521760687129e-05,
"loss": 0.1814,
"step": 1950
},
{
"epoch": 0.7852564102564102,
"grad_norm": 0.21755458414554596,
"learning_rate": 6.930637496352975e-05,
"loss": 0.1869,
"step": 1960
},
{
"epoch": 0.7892628205128205,
"grad_norm": 0.19528847932815552,
"learning_rate": 6.90067195843919e-05,
"loss": 0.181,
"step": 1970
},
{
"epoch": 0.7932692307692307,
"grad_norm": 0.2116439789533615,
"learning_rate": 6.870626408397775e-05,
"loss": 0.1878,
"step": 1980
},
{
"epoch": 0.7972756410256411,
"grad_norm": 0.21799945831298828,
"learning_rate": 6.840502111048981e-05,
"loss": 0.1774,
"step": 1990
},
{
"epoch": 0.8012820512820513,
"grad_norm": 0.21898455917835236,
"learning_rate": 6.810300334528069e-05,
"loss": 0.1873,
"step": 2000
},
{
"epoch": 0.8052884615384616,
"grad_norm": 0.23046636581420898,
"learning_rate": 6.780022350231912e-05,
"loss": 0.1817,
"step": 2010
},
{
"epoch": 0.8092948717948718,
"grad_norm": 0.2145206779241562,
"learning_rate": 6.749669432765496e-05,
"loss": 0.1818,
"step": 2020
},
{
"epoch": 0.813301282051282,
"grad_norm": 0.21499067544937134,
"learning_rate": 6.719242859888244e-05,
"loss": 0.1819,
"step": 2030
},
{
"epoch": 0.8173076923076923,
"grad_norm": 0.23782077431678772,
"learning_rate": 6.688743912460229e-05,
"loss": 0.1813,
"step": 2040
},
{
"epoch": 0.8213141025641025,
"grad_norm": 0.2248157560825348,
"learning_rate": 6.65817387438827e-05,
"loss": 0.1879,
"step": 2050
},
{
"epoch": 0.8253205128205128,
"grad_norm": 0.25905272364616394,
"learning_rate": 6.627534032571859e-05,
"loss": 0.1895,
"step": 2060
},
{
"epoch": 0.8293269230769231,
"grad_norm": 0.21141692996025085,
"learning_rate": 6.596825676849006e-05,
"loss": 0.1812,
"step": 2070
},
{
"epoch": 0.8333333333333334,
"grad_norm": 0.2148115485906601,
"learning_rate": 6.566050099941934e-05,
"loss": 0.1766,
"step": 2080
},
{
"epoch": 0.8373397435897436,
"grad_norm": 0.24514557421207428,
"learning_rate": 6.535208597402658e-05,
"loss": 0.1805,
"step": 2090
},
{
"epoch": 0.8413461538461539,
"grad_norm": 0.20855912566184998,
"learning_rate": 6.50430246755845e-05,
"loss": 0.1908,
"step": 2100
},
{
"epoch": 0.8453525641025641,
"grad_norm": 0.21720488369464874,
"learning_rate": 6.47333301145718e-05,
"loss": 0.2313,
"step": 2110
},
{
"epoch": 0.8493589743589743,
"grad_norm": 0.22862492501735687,
"learning_rate": 6.442301532812552e-05,
"loss": 0.1867,
"step": 2120
},
{
"epoch": 0.8533653846153846,
"grad_norm": 0.23019257187843323,
"learning_rate": 6.411209337949214e-05,
"loss": 0.1857,
"step": 2130
},
{
"epoch": 0.8573717948717948,
"grad_norm": 0.21849659085273743,
"learning_rate": 6.380057735747768e-05,
"loss": 0.1795,
"step": 2140
},
{
"epoch": 0.8613782051282052,
"grad_norm": 0.22794140875339508,
"learning_rate": 6.348848037589683e-05,
"loss": 0.1796,
"step": 2150
},
{
"epoch": 0.8653846153846154,
"grad_norm": 0.22126458585262299,
"learning_rate": 6.317581557302066e-05,
"loss": 0.1854,
"step": 2160
},
{
"epoch": 0.8693910256410257,
"grad_norm": 0.24870412051677704,
"learning_rate": 6.286259611102377e-05,
"loss": 0.1771,
"step": 2170
},
{
"epoch": 0.8733974358974359,
"grad_norm": 0.226875439286232,
"learning_rate": 6.254883517543006e-05,
"loss": 0.1769,
"step": 2180
},
{
"epoch": 0.8774038461538461,
"grad_norm": 0.2393677532672882,
"learning_rate": 6.223454597455776e-05,
"loss": 0.1878,
"step": 2190
},
{
"epoch": 0.8814102564102564,
"grad_norm": 0.23450277745723724,
"learning_rate": 6.191974173896329e-05,
"loss": 0.1795,
"step": 2200
},
{
"epoch": 0.8854166666666666,
"grad_norm": 0.23429277539253235,
"learning_rate": 6.160443572088444e-05,
"loss": 0.1809,
"step": 2210
},
{
"epoch": 0.8894230769230769,
"grad_norm": 0.2641499638557434,
"learning_rate": 6.128864119368234e-05,
"loss": 0.1908,
"step": 2220
},
{
"epoch": 0.8934294871794872,
"grad_norm": 0.25299501419067383,
"learning_rate": 6.0972371451282817e-05,
"loss": 0.1935,
"step": 2230
},
{
"epoch": 0.8974358974358975,
"grad_norm": 0.21204617619514465,
"learning_rate": 6.0655639807616705e-05,
"loss": 0.1871,
"step": 2240
},
{
"epoch": 0.9014423076923077,
"grad_norm": 0.20044928789138794,
"learning_rate": 6.0338459596059395e-05,
"loss": 0.185,
"step": 2250
},
{
"epoch": 0.905448717948718,
"grad_norm": 0.22311033308506012,
"learning_rate": 6.0020844168869496e-05,
"loss": 0.1809,
"step": 2260
},
{
"epoch": 0.9094551282051282,
"grad_norm": 0.2073046863079071,
"learning_rate": 5.970280689662685e-05,
"loss": 0.1822,
"step": 2270
},
{
"epoch": 0.9134615384615384,
"grad_norm": 0.23054327070713043,
"learning_rate": 5.938436116766954e-05,
"loss": 0.1862,
"step": 2280
},
{
"epoch": 0.9174679487179487,
"grad_norm": 0.2098427712917328,
"learning_rate": 5.9065520387530394e-05,
"loss": 0.1766,
"step": 2290
},
{
"epoch": 0.9214743589743589,
"grad_norm": 0.2541731595993042,
"learning_rate": 5.874629797837259e-05,
"loss": 0.1779,
"step": 2300
},
{
"epoch": 0.9254807692307693,
"grad_norm": 0.22261784970760345,
"learning_rate": 5.8426707378424675e-05,
"loss": 0.1885,
"step": 2310
},
{
"epoch": 0.9294871794871795,
"grad_norm": 0.23702876269817352,
"learning_rate": 5.8106762041414795e-05,
"loss": 0.1808,
"step": 2320
},
{
"epoch": 0.9334935897435898,
"grad_norm": 0.22513940930366516,
"learning_rate": 5.778647543600443e-05,
"loss": 0.1861,
"step": 2330
},
{
"epoch": 0.9375,
"grad_norm": 0.2372877299785614,
"learning_rate": 5.7465861045221295e-05,
"loss": 0.1831,
"step": 2340
},
{
"epoch": 0.9415064102564102,
"grad_norm": 0.2384011596441269,
"learning_rate": 5.714493236589187e-05,
"loss": 0.1835,
"step": 2350
},
{
"epoch": 0.9455128205128205,
"grad_norm": 0.22022205591201782,
"learning_rate": 5.6823702908073104e-05,
"loss": 0.1836,
"step": 2360
},
{
"epoch": 0.9495192307692307,
"grad_norm": 0.2236844301223755,
"learning_rate": 5.6502186194483783e-05,
"loss": 0.1797,
"step": 2370
},
{
"epoch": 0.9535256410256411,
"grad_norm": 0.24858827888965607,
"learning_rate": 5.6180395759935234e-05,
"loss": 0.1947,
"step": 2380
},
{
"epoch": 0.9575320512820513,
"grad_norm": 0.19967874884605408,
"learning_rate": 5.5858345150761514e-05,
"loss": 0.1831,
"step": 2390
},
{
"epoch": 0.9615384615384616,
"grad_norm": 0.25792399048805237,
"learning_rate": 5.553604792424922e-05,
"loss": 0.1875,
"step": 2400
},
{
"epoch": 0.9655448717948718,
"grad_norm": 0.22758032381534576,
"learning_rate": 5.521351764806671e-05,
"loss": 0.1851,
"step": 2410
},
{
"epoch": 0.969551282051282,
"grad_norm": 0.20984916388988495,
"learning_rate": 5.4890767899693015e-05,
"loss": 0.1811,
"step": 2420
},
{
"epoch": 0.9735576923076923,
"grad_norm": 0.2236003577709198,
"learning_rate": 5.4567812265846206e-05,
"loss": 0.1824,
"step": 2430
},
{
"epoch": 0.9775641025641025,
"grad_norm": 0.2495725303888321,
"learning_rate": 5.424466434191146e-05,
"loss": 0.1838,
"step": 2440
},
{
"epoch": 0.9815705128205128,
"grad_norm": 0.21274955570697784,
"learning_rate": 5.392133773136877e-05,
"loss": 0.1818,
"step": 2450
},
{
"epoch": 0.9855769230769231,
"grad_norm": 0.24202273786067963,
"learning_rate": 5.359784604522023e-05,
"loss": 0.1821,
"step": 2460
},
{
"epoch": 0.9895833333333334,
"grad_norm": 0.21269406378269196,
"learning_rate": 5.3274202901417115e-05,
"loss": 0.1853,
"step": 2470
},
{
"epoch": 0.9935897435897436,
"grad_norm": 0.2370026856660843,
"learning_rate": 5.2950421924286544e-05,
"loss": 0.167,
"step": 2480
},
{
"epoch": 0.9975961538461539,
"grad_norm": 0.1984049528837204,
"learning_rate": 5.262651674395799e-05,
"loss": 0.1715,
"step": 2490
},
{
"epoch": 1.0,
"eval_loss": 0.18003696203231812,
"eval_runtime": 312.955,
"eval_samples_per_second": 26.863,
"eval_steps_per_second": 3.358,
"step": 2496
},
{
"epoch": 1.001602564102564,
"grad_norm": 0.2148449867963791,
"learning_rate": 5.230250099578945e-05,
"loss": 0.1671,
"step": 2500
},
{
"epoch": 1.0056089743589745,
"grad_norm": 0.23484507203102112,
"learning_rate": 5.197838831979351e-05,
"loss": 0.1478,
"step": 2510
},
{
"epoch": 1.0096153846153846,
"grad_norm": 0.23629869520664215,
"learning_rate": 5.165419236006308e-05,
"loss": 0.1466,
"step": 2520
},
{
"epoch": 1.013621794871795,
"grad_norm": 0.2193659245967865,
"learning_rate": 5.132992676419703e-05,
"loss": 0.1487,
"step": 2530
},
{
"epoch": 1.017628205128205,
"grad_norm": 0.23937292397022247,
"learning_rate": 5.100560518272573e-05,
"loss": 0.1453,
"step": 2540
},
{
"epoch": 1.0216346153846154,
"grad_norm": 0.22235561907291412,
"learning_rate": 5.068124126853633e-05,
"loss": 0.1497,
"step": 2550
},
{
"epoch": 1.0256410256410255,
"grad_norm": 0.22830457985401154,
"learning_rate": 5.035684867629803e-05,
"loss": 0.1532,
"step": 2560
},
{
"epoch": 1.029647435897436,
"grad_norm": 0.23195677995681763,
"learning_rate": 5.003244106188735e-05,
"loss": 0.1531,
"step": 2570
},
{
"epoch": 1.0336538461538463,
"grad_norm": 0.24194592237472534,
"learning_rate": 4.9708032081813144e-05,
"loss": 0.1543,
"step": 2580
},
{
"epoch": 1.0376602564102564,
"grad_norm": 0.24159836769104004,
"learning_rate": 4.938363539264175e-05,
"loss": 0.1569,
"step": 2590
},
{
"epoch": 1.0416666666666667,
"grad_norm": 0.24700869619846344,
"learning_rate": 4.90592646504221e-05,
"loss": 0.1502,
"step": 2600
},
{
"epoch": 1.0456730769230769,
"grad_norm": 0.24695070087909698,
"learning_rate": 4.8734933510110905e-05,
"loss": 0.1555,
"step": 2610
},
{
"epoch": 1.0496794871794872,
"grad_norm": 0.21207590401172638,
"learning_rate": 4.841065562499769e-05,
"loss": 0.1495,
"step": 2620
},
{
"epoch": 1.0536858974358974,
"grad_norm": 0.22604689002037048,
"learning_rate": 4.8086444646130144e-05,
"loss": 0.1505,
"step": 2630
},
{
"epoch": 1.0576923076923077,
"grad_norm": 0.23026780784130096,
"learning_rate": 4.7762314221739416e-05,
"loss": 0.1414,
"step": 2640
},
{
"epoch": 1.061698717948718,
"grad_norm": 0.2474188357591629,
"learning_rate": 4.7438277996665546e-05,
"loss": 0.1563,
"step": 2650
},
{
"epoch": 1.0657051282051282,
"grad_norm": 0.2669748067855835,
"learning_rate": 4.7114349611783124e-05,
"loss": 0.1446,
"step": 2660
},
{
"epoch": 1.0697115384615385,
"grad_norm": 0.2164994478225708,
"learning_rate": 4.679054270342703e-05,
"loss": 0.1557,
"step": 2670
},
{
"epoch": 1.0737179487179487,
"grad_norm": 0.2127254158258438,
"learning_rate": 4.646687090281832e-05,
"loss": 0.1489,
"step": 2680
},
{
"epoch": 1.077724358974359,
"grad_norm": 0.25909334421157837,
"learning_rate": 4.614334783549049e-05,
"loss": 0.1517,
"step": 2690
},
{
"epoch": 1.0817307692307692,
"grad_norm": 0.20796196162700653,
"learning_rate": 4.581998712071583e-05,
"loss": 0.1459,
"step": 2700
},
{
"epoch": 1.0857371794871795,
"grad_norm": 0.21044594049453735,
"learning_rate": 4.549680237093215e-05,
"loss": 0.1508,
"step": 2710
},
{
"epoch": 1.0897435897435896,
"grad_norm": 0.22035610675811768,
"learning_rate": 4.517380719116966e-05,
"loss": 0.1518,
"step": 2720
},
{
"epoch": 1.09375,
"grad_norm": 0.2175433337688446,
"learning_rate": 4.485101517847831e-05,
"loss": 0.1532,
"step": 2730
},
{
"epoch": 1.0977564102564104,
"grad_norm": 0.23213978111743927,
"learning_rate": 4.452843992135536e-05,
"loss": 0.1503,
"step": 2740
},
{
"epoch": 1.1017628205128205,
"grad_norm": 0.22997038066387177,
"learning_rate": 4.420609499917337e-05,
"loss": 0.1532,
"step": 2750
},
{
"epoch": 1.1057692307692308,
"grad_norm": 0.24814608693122864,
"learning_rate": 4.3883993981608576e-05,
"loss": 0.1541,
"step": 2760
},
{
"epoch": 1.109775641025641,
"grad_norm": 0.24005189538002014,
"learning_rate": 4.3562150428069565e-05,
"loss": 0.1493,
"step": 2770
},
{
"epoch": 1.1137820512820513,
"grad_norm": 0.24364899098873138,
"learning_rate": 4.3240577887126565e-05,
"loss": 0.1507,
"step": 2780
},
{
"epoch": 1.1177884615384615,
"grad_norm": 0.25002944469451904,
"learning_rate": 4.291928989594101e-05,
"loss": 0.1493,
"step": 2790
},
{
"epoch": 1.1217948717948718,
"grad_norm": 0.24269698560237885,
"learning_rate": 4.2598299979695795e-05,
"loss": 0.146,
"step": 2800
},
{
"epoch": 1.125801282051282,
"grad_norm": 0.2566970884799957,
"learning_rate": 4.2277621651025736e-05,
"loss": 0.1498,
"step": 2810
},
{
"epoch": 1.1298076923076923,
"grad_norm": 0.24423649907112122,
"learning_rate": 4.195726840944886e-05,
"loss": 0.1562,
"step": 2820
},
{
"epoch": 1.1338141025641026,
"grad_norm": 0.2457042932510376,
"learning_rate": 4.1637253740798075e-05,
"loss": 0.1532,
"step": 2830
},
{
"epoch": 1.1378205128205128,
"grad_norm": 0.2585982382297516,
"learning_rate": 4.131759111665349e-05,
"loss": 0.1487,
"step": 2840
},
{
"epoch": 1.1418269230769231,
"grad_norm": 0.2370419055223465,
"learning_rate": 4.0998293993775237e-05,
"loss": 0.1442,
"step": 2850
},
{
"epoch": 1.1458333333333333,
"grad_norm": 0.25113531947135925,
"learning_rate": 4.067937581353708e-05,
"loss": 0.1459,
"step": 2860
},
{
"epoch": 1.1498397435897436,
"grad_norm": 0.22391347587108612,
"learning_rate": 4.036085000136046e-05,
"loss": 0.149,
"step": 2870
},
{
"epoch": 1.1538461538461537,
"grad_norm": 0.2450188249349594,
"learning_rate": 4.004272996614948e-05,
"loss": 0.1402,
"step": 2880
},
{
"epoch": 1.157852564102564,
"grad_norm": 0.2271914780139923,
"learning_rate": 3.972502909972629e-05,
"loss": 0.1488,
"step": 2890
},
{
"epoch": 1.1618589743589745,
"grad_norm": 0.2330724447965622,
"learning_rate": 3.940776077626739e-05,
"loss": 0.1517,
"step": 2900
},
{
"epoch": 1.1658653846153846,
"grad_norm": 0.23422060906887054,
"learning_rate": 3.909093835174066e-05,
"loss": 0.1475,
"step": 2910
},
{
"epoch": 1.169871794871795,
"grad_norm": 0.2604200541973114,
"learning_rate": 3.8774575163343065e-05,
"loss": 0.1531,
"step": 2920
},
{
"epoch": 1.173878205128205,
"grad_norm": 0.23886078596115112,
"learning_rate": 3.8458684528939185e-05,
"loss": 0.142,
"step": 2930
},
{
"epoch": 1.1778846153846154,
"grad_norm": 0.24594782292842865,
"learning_rate": 3.814327974650067e-05,
"loss": 0.1477,
"step": 2940
},
{
"epoch": 1.1818910256410255,
"grad_norm": 0.247484028339386,
"learning_rate": 3.7828374093546306e-05,
"loss": 0.1516,
"step": 2950
},
{
"epoch": 1.185897435897436,
"grad_norm": 0.22961559891700745,
"learning_rate": 3.751398082658322e-05,
"loss": 0.1508,
"step": 2960
},
{
"epoch": 1.1899038461538463,
"grad_norm": 0.2492343634366989,
"learning_rate": 3.720011318054871e-05,
"loss": 0.1412,
"step": 2970
},
{
"epoch": 1.1939102564102564,
"grad_norm": 0.24057136476039886,
"learning_rate": 3.688678436825313e-05,
"loss": 0.1476,
"step": 2980
},
{
"epoch": 1.1979166666666667,
"grad_norm": 0.2430902123451233,
"learning_rate": 3.657400757982366e-05,
"loss": 0.1506,
"step": 2990
},
{
"epoch": 1.2019230769230769,
"grad_norm": 0.23617245256900787,
"learning_rate": 3.626179598214917e-05,
"loss": 0.1454,
"step": 3000
},
{
"epoch": 1.2059294871794872,
"grad_norm": 0.26724281907081604,
"learning_rate": 3.595016271832572e-05,
"loss": 0.1435,
"step": 3010
},
{
"epoch": 1.2099358974358974,
"grad_norm": 0.24820230901241302,
"learning_rate": 3.5639120907103416e-05,
"loss": 0.1435,
"step": 3020
},
{
"epoch": 1.2139423076923077,
"grad_norm": 0.2463800460100174,
"learning_rate": 3.532868364233416e-05,
"loss": 0.1529,
"step": 3030
},
{
"epoch": 1.217948717948718,
"grad_norm": 0.25976675748825073,
"learning_rate": 3.501886399242038e-05,
"loss": 0.15,
"step": 3040
},
{
"epoch": 1.2219551282051282,
"grad_norm": 0.22920717298984528,
"learning_rate": 3.470967499976492e-05,
"loss": 0.1449,
"step": 3050
},
{
"epoch": 1.2259615384615385,
"grad_norm": 0.2552580237388611,
"learning_rate": 3.440112968022201e-05,
"loss": 0.1453,
"step": 3060
},
{
"epoch": 1.2299679487179487,
"grad_norm": 0.22986732423305511,
"learning_rate": 3.409324102254932e-05,
"loss": 0.1434,
"step": 3070
},
{
"epoch": 1.233974358974359,
"grad_norm": 0.2583804130554199,
"learning_rate": 3.37860219878612e-05,
"loss": 0.1532,
"step": 3080
},
{
"epoch": 1.2379807692307692,
"grad_norm": 0.24155107140541077,
"learning_rate": 3.347948550908303e-05,
"loss": 0.1482,
"step": 3090
},
{
"epoch": 1.2419871794871795,
"grad_norm": 0.25099289417266846,
"learning_rate": 3.31736444904068e-05,
"loss": 0.1494,
"step": 3100
},
{
"epoch": 1.2459935897435896,
"grad_norm": 0.26350319385528564,
"learning_rate": 3.286851180674788e-05,
"loss": 0.1441,
"step": 3110
},
{
"epoch": 1.25,
"grad_norm": 0.23903068900108337,
"learning_rate": 3.2564100303203035e-05,
"loss": 0.1455,
"step": 3120
},
{
"epoch": 1.2540064102564101,
"grad_norm": 0.25470852851867676,
"learning_rate": 3.22604227945097e-05,
"loss": 0.149,
"step": 3130
},
{
"epoch": 1.2580128205128205,
"grad_norm": 0.25256025791168213,
"learning_rate": 3.195749206450649e-05,
"loss": 0.1484,
"step": 3140
},
{
"epoch": 1.2620192307692308,
"grad_norm": 0.2604450583457947,
"learning_rate": 3.165532086559504e-05,
"loss": 0.1438,
"step": 3150
},
{
"epoch": 1.266025641025641,
"grad_norm": 0.2456076741218567,
"learning_rate": 3.135392191820324e-05,
"loss": 0.1475,
"step": 3160
},
{
"epoch": 1.2700320512820513,
"grad_norm": 0.23290812969207764,
"learning_rate": 3.1053307910249654e-05,
"loss": 0.1473,
"step": 3170
},
{
"epoch": 1.2740384615384617,
"grad_norm": 0.2368566393852234,
"learning_rate": 3.0753491496609445e-05,
"loss": 0.1468,
"step": 3180
},
{
"epoch": 1.2780448717948718,
"grad_norm": 0.242570161819458,
"learning_rate": 3.045448529858165e-05,
"loss": 0.1489,
"step": 3190
},
{
"epoch": 1.282051282051282,
"grad_norm": 0.23196978867053986,
"learning_rate": 3.0156301903357864e-05,
"loss": 0.1479,
"step": 3200
},
{
"epoch": 1.2860576923076923,
"grad_norm": 0.2525078356266022,
"learning_rate": 2.9858953863492334e-05,
"loss": 0.1508,
"step": 3210
},
{
"epoch": 1.2900641025641026,
"grad_norm": 0.23984560370445251,
"learning_rate": 2.956245369637356e-05,
"loss": 0.1428,
"step": 3220
},
{
"epoch": 1.2940705128205128,
"grad_norm": 0.2553931176662445,
"learning_rate": 2.9266813883697342e-05,
"loss": 0.1542,
"step": 3230
},
{
"epoch": 1.2980769230769231,
"grad_norm": 0.2719983458518982,
"learning_rate": 2.897204687094136e-05,
"loss": 0.1533,
"step": 3240
},
{
"epoch": 1.3020833333333333,
"grad_norm": 0.26820674538612366,
"learning_rate": 2.8678165066841257e-05,
"loss": 0.144,
"step": 3250
},
{
"epoch": 1.3060897435897436,
"grad_norm": 0.22624927759170532,
"learning_rate": 2.8385180842868207e-05,
"loss": 0.1453,
"step": 3260
},
{
"epoch": 1.3100961538461537,
"grad_norm": 0.2596217095851898,
"learning_rate": 2.8093106532708253e-05,
"loss": 0.1365,
"step": 3270
},
{
"epoch": 1.314102564102564,
"grad_norm": 0.24350501596927643,
"learning_rate": 2.780195443174293e-05,
"loss": 0.1458,
"step": 3280
},
{
"epoch": 1.3181089743589745,
"grad_norm": 0.2782920300960541,
"learning_rate": 2.7511736796531838e-05,
"loss": 0.1437,
"step": 3290
},
{
"epoch": 1.3221153846153846,
"grad_norm": 0.2512044310569763,
"learning_rate": 2.722246584429652e-05,
"loss": 0.1473,
"step": 3300
},
{
"epoch": 1.326121794871795,
"grad_norm": 0.24395060539245605,
"learning_rate": 2.6934153752406287e-05,
"loss": 0.1437,
"step": 3310
},
{
"epoch": 1.330128205128205,
"grad_norm": 0.24087293446063995,
"learning_rate": 2.6646812657865557e-05,
"loss": 0.1465,
"step": 3320
},
{
"epoch": 1.3341346153846154,
"grad_norm": 0.23852036893367767,
"learning_rate": 2.636045465680282e-05,
"loss": 0.1386,
"step": 3330
},
{
"epoch": 1.3381410256410255,
"grad_norm": 0.2611803114414215,
"learning_rate": 2.6075091803961616e-05,
"loss": 0.1436,
"step": 3340
},
{
"epoch": 1.342147435897436,
"grad_norm": 0.2321190983057022,
"learning_rate": 2.5790736112192894e-05,
"loss": 0.1507,
"step": 3350
},
{
"epoch": 1.3461538461538463,
"grad_norm": 0.2473578155040741,
"learning_rate": 2.550739955194944e-05,
"loss": 0.1422,
"step": 3360
},
{
"epoch": 1.3501602564102564,
"grad_norm": 0.2508649528026581,
"learning_rate": 2.5225094050781872e-05,
"loss": 0.1432,
"step": 3370
},
{
"epoch": 1.3541666666666667,
"grad_norm": 0.2570420205593109,
"learning_rate": 2.4943831492836588e-05,
"loss": 0.1531,
"step": 3380
},
{
"epoch": 1.3581730769230769,
"grad_norm": 0.24803844094276428,
"learning_rate": 2.4663623718355444e-05,
"loss": 0.2464,
"step": 3390
},
{
"epoch": 1.3621794871794872,
"grad_norm": 0.23173387348651886,
"learning_rate": 2.438448252317736e-05,
"loss": 0.1485,
"step": 3400
},
{
"epoch": 1.3661858974358974,
"grad_norm": 0.23720696568489075,
"learning_rate": 2.410641965824166e-05,
"loss": 0.1428,
"step": 3410
},
{
"epoch": 1.3701923076923077,
"grad_norm": 0.2676253318786621,
"learning_rate": 2.3829446829093537e-05,
"loss": 0.1471,
"step": 3420
},
{
"epoch": 1.374198717948718,
"grad_norm": 0.2445019632577896,
"learning_rate": 2.355357569539114e-05,
"loss": 0.1392,
"step": 3430
},
{
"epoch": 1.3782051282051282,
"grad_norm": 0.22741925716400146,
"learning_rate": 2.3278817870414866e-05,
"loss": 0.15,
"step": 3440
},
{
"epoch": 1.3822115384615383,
"grad_norm": 0.24508029222488403,
"learning_rate": 2.300518492057842e-05,
"loss": 0.1492,
"step": 3450
},
{
"epoch": 1.3862179487179487,
"grad_norm": 0.25865378975868225,
"learning_rate": 2.2732688364941862e-05,
"loss": 0.1494,
"step": 3460
},
{
"epoch": 1.390224358974359,
"grad_norm": 0.2557166814804077,
"learning_rate": 2.2461339674726806e-05,
"loss": 0.1486,
"step": 3470
},
{
"epoch": 1.3942307692307692,
"grad_norm": 0.25431498885154724,
"learning_rate": 2.219115027283339e-05,
"loss": 0.1443,
"step": 3480
},
{
"epoch": 1.3982371794871795,
"grad_norm": 0.2517376244068146,
"learning_rate": 2.192213153335953e-05,
"loss": 0.1434,
"step": 3490
},
{
"epoch": 1.4022435897435899,
"grad_norm": 0.24723024666309357,
"learning_rate": 2.165429478112199e-05,
"loss": 0.1405,
"step": 3500
},
{
"epoch": 1.40625,
"grad_norm": 0.24276477098464966,
"learning_rate": 2.138765129117977e-05,
"loss": 0.1483,
"step": 3510
},
{
"epoch": 1.4102564102564101,
"grad_norm": 0.23349566757678986,
"learning_rate": 2.1122212288359335e-05,
"loss": 0.1508,
"step": 3520
},
{
"epoch": 1.4142628205128205,
"grad_norm": 0.2291381061077118,
"learning_rate": 2.085798894678217e-05,
"loss": 0.1502,
"step": 3530
},
{
"epoch": 1.4182692307692308,
"grad_norm": 0.2317928522825241,
"learning_rate": 2.0594992389394395e-05,
"loss": 0.1408,
"step": 3540
},
{
"epoch": 1.422275641025641,
"grad_norm": 0.2636735141277313,
"learning_rate": 2.0333233687498433e-05,
"loss": 0.1418,
"step": 3550
},
{
"epoch": 1.4262820512820513,
"grad_norm": 0.2535155415534973,
"learning_rate": 2.0072723860287056e-05,
"loss": 0.1484,
"step": 3560
},
{
"epoch": 1.4302884615384617,
"grad_norm": 0.2632095217704773,
"learning_rate": 1.9813473874379395e-05,
"loss": 0.147,
"step": 3570
},
{
"epoch": 1.4342948717948718,
"grad_norm": 0.25649628043174744,
"learning_rate": 1.9555494643359457e-05,
"loss": 0.1464,
"step": 3580
},
{
"epoch": 1.438301282051282,
"grad_norm": 0.2731545865535736,
"learning_rate": 1.9298797027316474e-05,
"loss": 0.1516,
"step": 3590
},
{
"epoch": 1.4423076923076923,
"grad_norm": 0.22635942697525024,
"learning_rate": 1.9043391832387914e-05,
"loss": 0.1405,
"step": 3600
},
{
"epoch": 1.4463141025641026,
"grad_norm": 0.24070985615253448,
"learning_rate": 1.878928981030445e-05,
"loss": 0.1427,
"step": 3610
},
{
"epoch": 1.4503205128205128,
"grad_norm": 0.29021933674812317,
"learning_rate": 1.853650165793747e-05,
"loss": 0.1435,
"step": 3620
},
{
"epoch": 1.4543269230769231,
"grad_norm": 0.2542175352573395,
"learning_rate": 1.8285038016848636e-05,
"loss": 0.1531,
"step": 3630
},
{
"epoch": 1.4583333333333333,
"grad_norm": 0.24497313797473907,
"learning_rate": 1.803490947284204e-05,
"loss": 0.1451,
"step": 3640
},
{
"epoch": 1.4623397435897436,
"grad_norm": 0.24667443335056305,
"learning_rate": 1.7786126555518485e-05,
"loss": 0.1417,
"step": 3650
},
{
"epoch": 1.4663461538461537,
"grad_norm": 0.2563740015029907,
"learning_rate": 1.7538699737832236e-05,
"loss": 0.1524,
"step": 3660
},
{
"epoch": 1.470352564102564,
"grad_norm": 0.2556226849555969,
"learning_rate": 1.7292639435650222e-05,
"loss": 0.141,
"step": 3670
},
{
"epoch": 1.4743589743589745,
"grad_norm": 0.23436102271080017,
"learning_rate": 1.704795600731341e-05,
"loss": 0.1477,
"step": 3680
},
{
"epoch": 1.4783653846153846,
"grad_norm": 0.2800445556640625,
"learning_rate": 1.6804659753200924e-05,
"loss": 0.1429,
"step": 3690
},
{
"epoch": 1.482371794871795,
"grad_norm": 0.28161168098449707,
"learning_rate": 1.6562760915296295e-05,
"loss": 0.1423,
"step": 3700
},
{
"epoch": 1.486378205128205,
"grad_norm": 0.2788082957267761,
"learning_rate": 1.63222696767564e-05,
"loss": 0.1465,
"step": 3710
},
{
"epoch": 1.4903846153846154,
"grad_norm": 0.2512800097465515,
"learning_rate": 1.608319616148271e-05,
"loss": 0.1354,
"step": 3720
},
{
"epoch": 1.4943910256410255,
"grad_norm": 0.24340157210826874,
"learning_rate": 1.5845550433695173e-05,
"loss": 0.1461,
"step": 3730
},
{
"epoch": 1.498397435897436,
"grad_norm": 0.25012919306755066,
"learning_rate": 1.560934249750851e-05,
"loss": 0.1493,
"step": 3740
},
{
"epoch": 1.5024038461538463,
"grad_norm": 0.242750346660614,
"learning_rate": 1.5374582296511053e-05,
"loss": 0.1411,
"step": 3750
},
{
"epoch": 1.5064102564102564,
"grad_norm": 0.23980247974395752,
"learning_rate": 1.5141279713346208e-05,
"loss": 0.1441,
"step": 3760
},
{
"epoch": 1.5104166666666665,
"grad_norm": 0.2511241137981415,
"learning_rate": 1.4909444569296333e-05,
"loss": 0.1447,
"step": 3770
},
{
"epoch": 1.5144230769230769,
"grad_norm": 0.24846923351287842,
"learning_rate": 1.4679086623869464e-05,
"loss": 0.1441,
"step": 3780
},
{
"epoch": 1.5184294871794872,
"grad_norm": 0.24027830362319946,
"learning_rate": 1.4450215574388265e-05,
"loss": 0.1402,
"step": 3790
},
{
"epoch": 1.5224358974358974,
"grad_norm": 0.2580472528934479,
"learning_rate": 1.4222841055581964e-05,
"loss": 0.1377,
"step": 3800
},
{
"epoch": 1.5264423076923077,
"grad_norm": 0.23837298154830933,
"learning_rate": 1.3996972639180644e-05,
"loss": 0.142,
"step": 3810
},
{
"epoch": 1.530448717948718,
"grad_norm": 0.22810089588165283,
"learning_rate": 1.3772619833512412e-05,
"loss": 0.1456,
"step": 3820
},
{
"epoch": 1.5344551282051282,
"grad_norm": 0.22962678968906403,
"learning_rate": 1.3549792083103036e-05,
"loss": 0.1374,
"step": 3830
},
{
"epoch": 1.5384615384615383,
"grad_norm": 0.2668929398059845,
"learning_rate": 1.332849876827842e-05,
"loss": 0.1499,
"step": 3840
},
{
"epoch": 1.5424679487179487,
"grad_norm": 0.24207815527915955,
"learning_rate": 1.3108749204769733e-05,
"loss": 0.1369,
"step": 3850
},
{
"epoch": 1.546474358974359,
"grad_norm": 0.2575761377811432,
"learning_rate": 1.2890552643321146e-05,
"loss": 0.1504,
"step": 3860
},
{
"epoch": 1.5504807692307692,
"grad_norm": 0.23695258796215057,
"learning_rate": 1.2673918269300556e-05,
"loss": 0.1422,
"step": 3870
},
{
"epoch": 1.5544871794871795,
"grad_norm": 0.2854330837726593,
"learning_rate": 1.2458855202312775e-05,
"loss": 0.1506,
"step": 3880
},
{
"epoch": 1.5584935897435899,
"grad_norm": 0.2639913260936737,
"learning_rate": 1.2245372495815726e-05,
"loss": 0.1386,
"step": 3890
},
{
"epoch": 1.5625,
"grad_norm": 0.26962000131607056,
"learning_rate": 1.203347913673924e-05,
"loss": 0.1469,
"step": 3900
},
{
"epoch": 1.5665064102564101,
"grad_norm": 0.2668505907058716,
"learning_rate": 1.1823184045106817e-05,
"loss": 0.1381,
"step": 3910
},
{
"epoch": 1.5705128205128205,
"grad_norm": 0.24999478459358215,
"learning_rate": 1.1614496073660026e-05,
"loss": 0.1488,
"step": 3920
},
{
"epoch": 1.5745192307692308,
"grad_norm": 0.2618054151535034,
"learning_rate": 1.1407424007485929e-05,
"loss": 0.1365,
"step": 3930
},
{
"epoch": 1.578525641025641,
"grad_norm": 0.25881797075271606,
"learning_rate": 1.120197656364722e-05,
"loss": 0.1424,
"step": 3940
},
{
"epoch": 1.5825320512820513,
"grad_norm": 0.23021487891674042,
"learning_rate": 1.0998162390815208e-05,
"loss": 0.1344,
"step": 3950
},
{
"epoch": 1.5865384615384617,
"grad_norm": 0.24012772738933563,
"learning_rate": 1.0795990068905843e-05,
"loss": 0.1511,
"step": 3960
},
{
"epoch": 1.5905448717948718,
"grad_norm": 0.251249760389328,
"learning_rate": 1.0595468108718449e-05,
"loss": 0.1433,
"step": 3970
},
{
"epoch": 1.594551282051282,
"grad_norm": 0.2358747273683548,
"learning_rate": 1.0396604951577483e-05,
"loss": 0.147,
"step": 3980
},
{
"epoch": 1.5985576923076923,
"grad_norm": 0.25139543414115906,
"learning_rate": 1.0199408968977136e-05,
"loss": 0.1478,
"step": 3990
},
{
"epoch": 1.6025641025641026,
"grad_norm": 0.25917524099349976,
"learning_rate": 1.0003888462229011e-05,
"loss": 0.1467,
"step": 4000
},
{
"epoch": 1.6065705128205128,
"grad_norm": 0.2669861316680908,
"learning_rate": 9.810051662112557e-06,
"loss": 0.1348,
"step": 4010
},
{
"epoch": 1.6105769230769231,
"grad_norm": 0.27329695224761963,
"learning_rate": 9.61790672852868e-06,
"loss": 0.1384,
"step": 4020
},
{
"epoch": 1.6145833333333335,
"grad_norm": 0.2671654522418976,
"learning_rate": 9.427461750156142e-06,
"loss": 0.1431,
"step": 4030
},
{
"epoch": 1.6185897435897436,
"grad_norm": 0.25406405329704285,
"learning_rate": 9.238724744111138e-06,
"loss": 0.1444,
"step": 4040
},
{
"epoch": 1.6225961538461537,
"grad_norm": 0.25970324873924255,
"learning_rate": 9.05170365560976e-06,
"loss": 0.1396,
"step": 4050
},
{
"epoch": 1.626602564102564,
"grad_norm": 0.2700863480567932,
"learning_rate": 8.86640635763351e-06,
"loss": 0.1484,
"step": 4060
},
{
"epoch": 1.6306089743589745,
"grad_norm": 0.24397385120391846,
"learning_rate": 8.68284065059794e-06,
"loss": 0.1407,
"step": 4070
},
{
"epoch": 1.6346153846153846,
"grad_norm": 0.2537296414375305,
"learning_rate": 8.501014262024176e-06,
"loss": 0.1468,
"step": 4080
},
{
"epoch": 1.6386217948717947,
"grad_norm": 0.2614051103591919,
"learning_rate": 8.320934846213746e-06,
"loss": 0.1368,
"step": 4090
},
{
"epoch": 1.6426282051282053,
"grad_norm": 0.2941496670246124,
"learning_rate": 8.142609983926225e-06,
"loss": 0.1434,
"step": 4100
},
{
"epoch": 1.6466346153846154,
"grad_norm": 0.2491438090801239,
"learning_rate": 7.966047182060226e-06,
"loss": 0.1433,
"step": 4110
},
{
"epoch": 1.6506410256410255,
"grad_norm": 0.267880916595459,
"learning_rate": 7.791253873337278e-06,
"loss": 0.1437,
"step": 4120
},
{
"epoch": 1.654647435897436,
"grad_norm": 0.25333133339881897,
"learning_rate": 7.618237415989032e-06,
"loss": 0.1433,
"step": 4130
},
{
"epoch": 1.6586538461538463,
"grad_norm": 0.25538668036460876,
"learning_rate": 7.44700509344744e-06,
"loss": 0.139,
"step": 4140
},
{
"epoch": 1.6626602564102564,
"grad_norm": 0.25621935725212097,
"learning_rate": 7.277564114038149e-06,
"loss": 0.1347,
"step": 4150
},
{
"epoch": 1.6666666666666665,
"grad_norm": 0.2569233775138855,
"learning_rate": 7.109921610677078e-06,
"loss": 0.1408,
"step": 4160
},
{
"epoch": 1.6706730769230769,
"grad_norm": 0.23728112876415253,
"learning_rate": 6.9440846405701424e-06,
"loss": 0.1352,
"step": 4170
},
{
"epoch": 1.6746794871794872,
"grad_norm": 0.23264577984809875,
"learning_rate": 6.780060184916159e-06,
"loss": 0.1463,
"step": 4180
},
{
"epoch": 1.6786858974358974,
"grad_norm": 0.24841400980949402,
"learning_rate": 6.6178551486129445e-06,
"loss": 0.1438,
"step": 4190
},
{
"epoch": 1.6826923076923077,
"grad_norm": 0.2776187062263489,
"learning_rate": 6.4574763599666856e-06,
"loss": 0.1438,
"step": 4200
},
{
"epoch": 1.686698717948718,
"grad_norm": 0.2587837874889374,
"learning_rate": 6.2989305704044325e-06,
"loss": 0.1367,
"step": 4210
},
{
"epoch": 1.6907051282051282,
"grad_norm": 0.2615884244441986,
"learning_rate": 6.142224454189954e-06,
"loss": 0.1726,
"step": 4220
},
{
"epoch": 1.6947115384615383,
"grad_norm": 0.25619083642959595,
"learning_rate": 5.987364608142693e-06,
"loss": 0.151,
"step": 4230
},
{
"epoch": 1.6987179487179487,
"grad_norm": 0.28548672795295715,
"learning_rate": 5.834357551360132e-06,
"loss": 0.1421,
"step": 4240
},
{
"epoch": 1.702724358974359,
"grad_norm": 0.26761943101882935,
"learning_rate": 5.683209724943345e-06,
"loss": 0.1383,
"step": 4250
},
{
"epoch": 1.7067307692307692,
"grad_norm": 0.2561447322368622,
"learning_rate": 5.533927491725787e-06,
"loss": 0.1421,
"step": 4260
},
{
"epoch": 1.7107371794871795,
"grad_norm": 0.2707176208496094,
"learning_rate": 5.3865171360055425e-06,
"loss": 0.1519,
"step": 4270
},
{
"epoch": 1.7147435897435899,
"grad_norm": 0.264593243598938,
"learning_rate": 5.240984863280668e-06,
"loss": 0.1464,
"step": 4280
},
{
"epoch": 1.71875,
"grad_norm": 0.2739621102809906,
"learning_rate": 5.097336799988067e-06,
"loss": 0.1392,
"step": 4290
},
{
"epoch": 1.7227564102564101,
"grad_norm": 0.23156005144119263,
"learning_rate": 4.955578993245485e-06,
"loss": 0.1377,
"step": 4300
},
{
"epoch": 1.7267628205128205,
"grad_norm": 0.24819669127464294,
"learning_rate": 4.815717410597043e-06,
"loss": 0.138,
"step": 4310
},
{
"epoch": 1.7307692307692308,
"grad_norm": 0.25722768902778625,
"learning_rate": 4.6777579397619285e-06,
"loss": 0.1434,
"step": 4320
},
{
"epoch": 1.734775641025641,
"grad_norm": 0.22565537691116333,
"learning_rate": 4.541706388386624e-06,
"loss": 0.1409,
"step": 4330
},
{
"epoch": 1.7387820512820513,
"grad_norm": 0.2539729177951813,
"learning_rate": 4.4075684838003784e-06,
"loss": 0.1447,
"step": 4340
},
{
"epoch": 1.7427884615384617,
"grad_norm": 0.2426980435848236,
"learning_rate": 4.275349872774098e-06,
"loss": 0.1422,
"step": 4350
},
{
"epoch": 1.7467948717948718,
"grad_norm": 0.26267197728157043,
"learning_rate": 4.14505612128267e-06,
"loss": 0.1381,
"step": 4360
},
{
"epoch": 1.750801282051282,
"grad_norm": 0.23471184074878693,
"learning_rate": 4.016692714270631e-06,
"loss": 0.1367,
"step": 4370
},
{
"epoch": 1.7548076923076923,
"grad_norm": 0.2703573703765869,
"learning_rate": 3.890265055421283e-06,
"loss": 0.144,
"step": 4380
},
{
"epoch": 1.7588141025641026,
"grad_norm": 0.23946279287338257,
"learning_rate": 3.7657784669291816e-06,
"loss": 0.1397,
"step": 4390
},
{
"epoch": 1.7628205128205128,
"grad_norm": 0.23393961787223816,
"learning_rate": 3.643238189276138e-06,
"loss": 0.1329,
"step": 4400
},
{
"epoch": 1.7668269230769231,
"grad_norm": 0.24087683856487274,
"learning_rate": 3.522649381010562e-06,
"loss": 0.1367,
"step": 4410
},
{
"epoch": 1.7708333333333335,
"grad_norm": 0.2636178135871887,
"learning_rate": 3.4040171185303616e-06,
"loss": 0.1415,
"step": 4420
},
{
"epoch": 1.7748397435897436,
"grad_norm": 0.2364189773797989,
"learning_rate": 3.2873463958691673e-06,
"loss": 0.1445,
"step": 4430
},
{
"epoch": 1.7788461538461537,
"grad_norm": 0.26020485162734985,
"learning_rate": 3.1726421244861917e-06,
"loss": 0.1388,
"step": 4440
},
{
"epoch": 1.782852564102564,
"grad_norm": 0.25162965059280396,
"learning_rate": 3.0599091330593797e-06,
"loss": 0.1427,
"step": 4450
},
{
"epoch": 1.7868589743589745,
"grad_norm": 0.24311134219169617,
"learning_rate": 2.949152167282204e-06,
"loss": 0.1313,
"step": 4460
},
{
"epoch": 1.7908653846153846,
"grad_norm": 0.249532088637352,
"learning_rate": 2.840375889663871e-06,
"loss": 0.1486,
"step": 4470
},
{
"epoch": 1.7948717948717947,
"grad_norm": 0.2393971085548401,
"learning_rate": 2.733584879333001e-06,
"loss": 0.1467,
"step": 4480
},
{
"epoch": 1.7988782051282053,
"grad_norm": 0.25224366784095764,
"learning_rate": 2.62878363184495e-06,
"loss": 0.1446,
"step": 4490
},
{
"epoch": 1.8028846153846154,
"grad_norm": 0.2636267840862274,
"learning_rate": 2.5259765589924544e-06,
"loss": 0.1487,
"step": 4500
},
{
"epoch": 1.8068910256410255,
"grad_norm": 0.25640687346458435,
"learning_rate": 2.425167988620014e-06,
"loss": 0.1358,
"step": 4510
},
{
"epoch": 1.810897435897436,
"grad_norm": 0.24088767170906067,
"learning_rate": 2.3263621644416034e-06,
"loss": 0.1404,
"step": 4520
},
{
"epoch": 1.8149038461538463,
"grad_norm": 0.2403586059808731,
"learning_rate": 2.2295632458621095e-06,
"loss": 0.1412,
"step": 4530
},
{
"epoch": 1.8189102564102564,
"grad_norm": 0.23450630903244019,
"learning_rate": 2.1347753078022005e-06,
"loss": 0.1447,
"step": 4540
},
{
"epoch": 1.8229166666666665,
"grad_norm": 0.2330087125301361,
"learning_rate": 2.0420023405267665e-06,
"loss": 0.1438,
"step": 4550
},
{
"epoch": 1.8269230769230769,
"grad_norm": 0.208469420671463,
"learning_rate": 1.9512482494769613e-06,
"loss": 0.1363,
"step": 4560
},
{
"epoch": 1.8309294871794872,
"grad_norm": 0.24098703265190125,
"learning_rate": 1.8625168551058114e-06,
"loss": 0.145,
"step": 4570
},
{
"epoch": 1.8349358974358974,
"grad_norm": 0.24833528697490692,
"learning_rate": 1.7758118927173606e-06,
"loss": 0.1417,
"step": 4580
},
{
"epoch": 1.8389423076923077,
"grad_norm": 0.2671266496181488,
"learning_rate": 1.6911370123094239e-06,
"loss": 0.141,
"step": 4590
},
{
"epoch": 1.842948717948718,
"grad_norm": 0.25512388348579407,
"learning_rate": 1.6084957784199662e-06,
"loss": 0.1357,
"step": 4600
},
{
"epoch": 1.8469551282051282,
"grad_norm": 0.2621639668941498,
"learning_rate": 1.5278916699770163e-06,
"loss": 0.1446,
"step": 4610
},
{
"epoch": 1.8509615384615383,
"grad_norm": 0.2648192048072815,
"learning_rate": 1.4493280801522347e-06,
"loss": 0.1391,
"step": 4620
},
{
"epoch": 1.8549679487179487,
"grad_norm": 0.26396840810775757,
"learning_rate": 1.3728083162180382e-06,
"loss": 0.1409,
"step": 4630
},
{
"epoch": 1.858974358974359,
"grad_norm": 0.24495120346546173,
"learning_rate": 1.2983355994084401e-06,
"loss": 0.1388,
"step": 4640
},
{
"epoch": 1.8629807692307692,
"grad_norm": 0.26375770568847656,
"learning_rate": 1.2259130647833627e-06,
"loss": 0.1376,
"step": 4650
},
{
"epoch": 1.8669871794871795,
"grad_norm": 0.25506845116615295,
"learning_rate": 1.155543761096739e-06,
"loss": 0.1432,
"step": 4660
},
{
"epoch": 1.8709935897435899,
"grad_norm": 0.2563297152519226,
"learning_rate": 1.0872306506681251e-06,
"loss": 0.143,
"step": 4670
},
{
"epoch": 1.875,
"grad_norm": 0.23788757622241974,
"learning_rate": 1.0209766092579897e-06,
"loss": 0.1387,
"step": 4680
},
{
"epoch": 1.8790064102564101,
"grad_norm": 0.2355082929134369,
"learning_rate": 9.56784425946705e-07,
"loss": 0.1403,
"step": 4690
},
{
"epoch": 1.8830128205128205,
"grad_norm": 0.26146796345710754,
"learning_rate": 8.946568030170688e-07,
"loss": 0.1413,
"step": 4700
},
{
"epoch": 1.8870192307692308,
"grad_norm": 0.2697630822658539,
"learning_rate": 8.345963558406e-07,
"loss": 0.147,
"step": 4710
},
{
"epoch": 1.891025641025641,
"grad_norm": 0.26238909363746643,
"learning_rate": 7.766056127674115e-07,
"loss": 0.1458,
"step": 4720
},
{
"epoch": 1.8950320512820513,
"grad_norm": 0.2590833902359009,
"learning_rate": 7.20687015019783e-07,
"loss": 0.1396,
"step": 4730
},
{
"epoch": 1.8990384615384617,
"grad_norm": 0.2681121230125427,
"learning_rate": 6.668429165893997e-07,
"loss": 0.1356,
"step": 4740
},
{
"epoch": 1.9030448717948718,
"grad_norm": 0.2643556296825409,
"learning_rate": 6.150755841382527e-07,
"loss": 0.1429,
"step": 4750
},
{
"epoch": 1.907051282051282,
"grad_norm": 0.2717660963535309,
"learning_rate": 5.653871969032054e-07,
"loss": 0.1409,
"step": 4760
},
{
"epoch": 1.9110576923076923,
"grad_norm": 0.24568721652030945,
"learning_rate": 5.177798466042716e-07,
"loss": 0.1444,
"step": 4770
},
{
"epoch": 1.9150641025641026,
"grad_norm": 0.23634164035320282,
"learning_rate": 4.722555373565751e-07,
"loss": 0.1472,
"step": 4780
},
{
"epoch": 1.9190705128205128,
"grad_norm": 0.23092865943908691,
"learning_rate": 4.288161855859285e-07,
"loss": 0.141,
"step": 4790
},
{
"epoch": 1.9230769230769231,
"grad_norm": 0.26985424757003784,
"learning_rate": 3.874636199482307e-07,
"loss": 0.1446,
"step": 4800
},
{
"epoch": 1.9270833333333335,
"grad_norm": 0.2507458031177521,
"learning_rate": 3.481995812524286e-07,
"loss": 0.1398,
"step": 4810
},
{
"epoch": 1.9310897435897436,
"grad_norm": 0.24660862982273102,
"learning_rate": 3.110257223872592e-07,
"loss": 0.1421,
"step": 4820
},
{
"epoch": 1.9350961538461537,
"grad_norm": 0.24846522510051727,
"learning_rate": 2.7594360825166644e-07,
"loss": 0.1439,
"step": 4830
},
{
"epoch": 1.939102564102564,
"grad_norm": 0.25589925050735474,
"learning_rate": 2.4295471568892005e-07,
"loss": 0.136,
"step": 4840
},
{
"epoch": 1.9431089743589745,
"grad_norm": 0.23860958218574524,
"learning_rate": 2.1206043342445469e-07,
"loss": 0.1446,
"step": 4850
},
{
"epoch": 1.9471153846153846,
"grad_norm": 0.2679506540298462,
"learning_rate": 1.8326206200739436e-07,
"loss": 0.1399,
"step": 4860
},
{
"epoch": 1.9511217948717947,
"grad_norm": 0.2246963083744049,
"learning_rate": 1.5656081375581277e-07,
"loss": 0.1435,
"step": 4870
},
{
"epoch": 1.9551282051282053,
"grad_norm": 0.23173221945762634,
"learning_rate": 1.3195781270570196e-07,
"loss": 0.1438,
"step": 4880
},
{
"epoch": 1.9591346153846154,
"grad_norm": 0.24970708787441254,
"learning_rate": 1.0945409456364352e-07,
"loss": 0.1429,
"step": 4890
},
{
"epoch": 1.9631410256410255,
"grad_norm": 0.23272265493869781,
"learning_rate": 8.90506066632102e-08,
"loss": 0.14,
"step": 4900
},
{
"epoch": 1.967147435897436,
"grad_norm": 0.25013232231140137,
"learning_rate": 7.074820792510317e-08,
"loss": 0.143,
"step": 4910
},
{
"epoch": 1.9711538461538463,
"grad_norm": 0.2563280165195465,
"learning_rate": 5.454766882097007e-08,
"loss": 0.149,
"step": 4920
},
{
"epoch": 1.9751602564102564,
"grad_norm": 0.25120726227760315,
"learning_rate": 4.0449671340991866e-08,
"loss": 0.1454,
"step": 4930
},
{
"epoch": 1.9791666666666665,
"grad_norm": 0.2662029564380646,
"learning_rate": 2.8454808965155954e-08,
"loss": 0.1461,
"step": 4940
},
{
"epoch": 1.9831730769230769,
"grad_norm": 0.24329765141010284,
"learning_rate": 1.8563586638281615e-08,
"loss": 0.1352,
"step": 4950
},
{
"epoch": 1.9871794871794872,
"grad_norm": 0.24964918196201324,
"learning_rate": 1.0776420748753734e-08,
"loss": 0.1378,
"step": 4960
},
{
"epoch": 1.9911858974358974,
"grad_norm": 0.24900883436203003,
"learning_rate": 5.093639111025672e-09,
"loss": 0.1451,
"step": 4970
},
{
"epoch": 1.9951923076923077,
"grad_norm": 0.24761377274990082,
"learning_rate": 1.5154809517747837e-09,
"loss": 0.1333,
"step": 4980
},
{
"epoch": 1.999198717948718,
"grad_norm": 0.2465759515762329,
"learning_rate": 4.2096899854904284e-11,
"loss": 0.1421,
"step": 4990
}
],
"logging_steps": 10,
"max_steps": 4992,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 1.3554923454948815e+19,
"train_batch_size": 64,
"trial_name": null,
"trial_params": null
}