bgem3-synthetic-e2 / trainer_state.json
nntoan209's picture
Upload folder using huggingface_hub
6eaf4df verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"eval_steps": 500,
"global_step": 14266,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"grad_norm": 9.326686626828339,
"learning_rate": 3.757943925233645e-07,
"loss": 0.4039,
"step": 20
},
{
"epoch": 0.01,
"grad_norm": 7.728208113985814,
"learning_rate": 5.608411214953272e-07,
"loss": 0.3994,
"step": 40
},
{
"epoch": 0.01,
"grad_norm": 4.700397296102845,
"learning_rate": 7.458878504672898e-07,
"loss": 0.39,
"step": 60
},
{
"epoch": 0.01,
"grad_norm": 4.700676061826897,
"learning_rate": 9.309345794392525e-07,
"loss": 0.3783,
"step": 80
},
{
"epoch": 0.01,
"grad_norm": 9.464699461085486,
"learning_rate": 1.115981308411215e-06,
"loss": 0.3765,
"step": 100
},
{
"epoch": 0.02,
"grad_norm": 4.415621915844531,
"learning_rate": 1.3010280373831777e-06,
"loss": 0.3618,
"step": 120
},
{
"epoch": 0.02,
"grad_norm": 5.642220142135915,
"learning_rate": 1.4860747663551404e-06,
"loss": 0.3458,
"step": 140
},
{
"epoch": 0.02,
"grad_norm": 4.258611593592227,
"learning_rate": 1.6711214953271029e-06,
"loss": 0.3166,
"step": 160
},
{
"epoch": 0.03,
"grad_norm": 5.262358877867667,
"learning_rate": 1.8561682242990656e-06,
"loss": 0.3329,
"step": 180
},
{
"epoch": 0.03,
"grad_norm": 5.572931858885786,
"learning_rate": 2.041214953271028e-06,
"loss": 0.3024,
"step": 200
},
{
"epoch": 0.03,
"grad_norm": 4.203318765137546,
"learning_rate": 2.2262616822429908e-06,
"loss": 0.3111,
"step": 220
},
{
"epoch": 0.03,
"grad_norm": 12.300229847318912,
"learning_rate": 2.4113084112149535e-06,
"loss": 0.2849,
"step": 240
},
{
"epoch": 0.04,
"grad_norm": 3.9064573919554753,
"learning_rate": 2.5963551401869158e-06,
"loss": 0.2824,
"step": 260
},
{
"epoch": 0.04,
"grad_norm": 4.020422062898716,
"learning_rate": 2.781401869158879e-06,
"loss": 0.2731,
"step": 280
},
{
"epoch": 0.04,
"grad_norm": 6.213285105658618,
"learning_rate": 2.9664485981308416e-06,
"loss": 0.2796,
"step": 300
},
{
"epoch": 0.04,
"grad_norm": 5.076365380233461,
"learning_rate": 3.1514953271028043e-06,
"loss": 0.252,
"step": 320
},
{
"epoch": 0.05,
"grad_norm": 4.492522278625916,
"learning_rate": 3.336542056074767e-06,
"loss": 0.2696,
"step": 340
},
{
"epoch": 0.05,
"grad_norm": 3.940312667678465,
"learning_rate": 3.5215887850467297e-06,
"loss": 0.2872,
"step": 360
},
{
"epoch": 0.05,
"grad_norm": 8.728447228208749,
"learning_rate": 3.7066355140186916e-06,
"loss": 0.2658,
"step": 380
},
{
"epoch": 0.06,
"grad_norm": 4.023556013121062,
"learning_rate": 3.891682242990654e-06,
"loss": 0.2674,
"step": 400
},
{
"epoch": 0.06,
"grad_norm": 4.313801223596667,
"learning_rate": 4.076728971962617e-06,
"loss": 0.2704,
"step": 420
},
{
"epoch": 0.06,
"grad_norm": 4.158184547794928,
"learning_rate": 4.26177570093458e-06,
"loss": 0.2532,
"step": 440
},
{
"epoch": 0.06,
"grad_norm": 4.298747150745207,
"learning_rate": 4.446822429906542e-06,
"loss": 0.2538,
"step": 460
},
{
"epoch": 0.07,
"grad_norm": 4.265066480576854,
"learning_rate": 4.631869158878505e-06,
"loss": 0.2386,
"step": 480
},
{
"epoch": 0.07,
"grad_norm": 4.553015688367492,
"learning_rate": 4.816915887850468e-06,
"loss": 0.2336,
"step": 500
},
{
"epoch": 0.07,
"grad_norm": 6.468109836322844,
"learning_rate": 5.00196261682243e-06,
"loss": 0.2412,
"step": 520
},
{
"epoch": 0.08,
"grad_norm": 4.900116603787392,
"learning_rate": 5.187009345794394e-06,
"loss": 0.2588,
"step": 540
},
{
"epoch": 0.08,
"grad_norm": 3.9237745327895026,
"learning_rate": 5.372056074766357e-06,
"loss": 0.2432,
"step": 560
},
{
"epoch": 0.08,
"grad_norm": 3.4008079242076454,
"learning_rate": 5.557102803738319e-06,
"loss": 0.2406,
"step": 580
},
{
"epoch": 0.08,
"grad_norm": 3.173130031399122,
"learning_rate": 5.742149532710281e-06,
"loss": 0.2313,
"step": 600
},
{
"epoch": 0.09,
"grad_norm": 4.517780779339123,
"learning_rate": 5.927196261682244e-06,
"loss": 0.2166,
"step": 620
},
{
"epoch": 0.09,
"grad_norm": 4.148419108951701,
"learning_rate": 6.112242990654207e-06,
"loss": 0.2578,
"step": 640
},
{
"epoch": 0.09,
"grad_norm": 3.3849149518055888,
"learning_rate": 6.297289719626169e-06,
"loss": 0.243,
"step": 660
},
{
"epoch": 0.1,
"grad_norm": 4.229179094167907,
"learning_rate": 6.482336448598131e-06,
"loss": 0.2442,
"step": 680
},
{
"epoch": 0.1,
"grad_norm": 4.898869302264945,
"learning_rate": 6.667383177570094e-06,
"loss": 0.2375,
"step": 700
},
{
"epoch": 0.1,
"grad_norm": 2.9552510999440815,
"learning_rate": 6.852429906542057e-06,
"loss": 0.2141,
"step": 720
},
{
"epoch": 0.1,
"grad_norm": 4.1613190221404945,
"learning_rate": 7.0374766355140186e-06,
"loss": 0.2369,
"step": 740
},
{
"epoch": 0.11,
"grad_norm": 3.586237211314309,
"learning_rate": 7.222523364485981e-06,
"loss": 0.2382,
"step": 760
},
{
"epoch": 0.11,
"grad_norm": 4.73959536149172,
"learning_rate": 7.407570093457944e-06,
"loss": 0.2211,
"step": 780
},
{
"epoch": 0.11,
"grad_norm": 3.5533342741010134,
"learning_rate": 7.592616822429907e-06,
"loss": 0.2535,
"step": 800
},
{
"epoch": 0.11,
"grad_norm": 4.12413246973587,
"learning_rate": 7.77766355140187e-06,
"loss": 0.2265,
"step": 820
},
{
"epoch": 0.12,
"grad_norm": 3.6092107826295914,
"learning_rate": 7.962710280373831e-06,
"loss": 0.2309,
"step": 840
},
{
"epoch": 0.12,
"grad_norm": 2.9958876398405394,
"learning_rate": 8.147757009345795e-06,
"loss": 0.215,
"step": 860
},
{
"epoch": 0.12,
"grad_norm": 3.7746974192230898,
"learning_rate": 8.332803738317757e-06,
"loss": 0.2367,
"step": 880
},
{
"epoch": 0.13,
"grad_norm": 4.765481515193955,
"learning_rate": 8.51785046728972e-06,
"loss": 0.2282,
"step": 900
},
{
"epoch": 0.13,
"grad_norm": 3.559142287151559,
"learning_rate": 8.702897196261682e-06,
"loss": 0.239,
"step": 920
},
{
"epoch": 0.13,
"grad_norm": 3.6955439876126333,
"learning_rate": 8.887943925233646e-06,
"loss": 0.212,
"step": 940
},
{
"epoch": 0.13,
"grad_norm": 3.2766607252541196,
"learning_rate": 9.072990654205607e-06,
"loss": 0.2189,
"step": 960
},
{
"epoch": 0.14,
"grad_norm": 3.933834004191608,
"learning_rate": 9.258037383177571e-06,
"loss": 0.2208,
"step": 980
},
{
"epoch": 0.14,
"grad_norm": 2.9425795445349525,
"learning_rate": 9.443084112149533e-06,
"loss": 0.2194,
"step": 1000
},
{
"epoch": 0.14,
"grad_norm": 2.959168562491463,
"learning_rate": 9.628130841121496e-06,
"loss": 0.2077,
"step": 1020
},
{
"epoch": 0.15,
"grad_norm": 5.0550970887351,
"learning_rate": 9.813177570093458e-06,
"loss": 0.2133,
"step": 1040
},
{
"epoch": 0.15,
"grad_norm": 3.676187652920463,
"learning_rate": 9.998224299065422e-06,
"loss": 0.2252,
"step": 1060
},
{
"epoch": 0.15,
"grad_norm": 3.2623851712068057,
"learning_rate": 1.0183271028037385e-05,
"loss": 0.2109,
"step": 1080
},
{
"epoch": 0.15,
"grad_norm": 2.790578221936117,
"learning_rate": 1.0368317757009347e-05,
"loss": 0.2184,
"step": 1100
},
{
"epoch": 0.16,
"grad_norm": 3.103974065416671,
"learning_rate": 1.055336448598131e-05,
"loss": 0.2236,
"step": 1120
},
{
"epoch": 0.16,
"grad_norm": 4.190626884306414,
"learning_rate": 1.0738411214953273e-05,
"loss": 0.2124,
"step": 1140
},
{
"epoch": 0.16,
"grad_norm": 3.5266419448659083,
"learning_rate": 1.0923457943925235e-05,
"loss": 0.2119,
"step": 1160
},
{
"epoch": 0.17,
"grad_norm": 4.268237533080129,
"learning_rate": 1.1108504672897196e-05,
"loss": 0.2132,
"step": 1180
},
{
"epoch": 0.17,
"grad_norm": 3.0890794428436474,
"learning_rate": 1.1293551401869158e-05,
"loss": 0.2144,
"step": 1200
},
{
"epoch": 0.17,
"grad_norm": 2.570523888148055,
"learning_rate": 1.147859813084112e-05,
"loss": 0.2249,
"step": 1220
},
{
"epoch": 0.17,
"grad_norm": 4.226701716529414,
"learning_rate": 1.1663644859813084e-05,
"loss": 0.2079,
"step": 1240
},
{
"epoch": 0.18,
"grad_norm": 3.1007684801088145,
"learning_rate": 1.1848691588785046e-05,
"loss": 0.2231,
"step": 1260
},
{
"epoch": 0.18,
"grad_norm": 3.1750343260824376,
"learning_rate": 1.2033738317757009e-05,
"loss": 0.213,
"step": 1280
},
{
"epoch": 0.18,
"grad_norm": 4.943978224548977,
"learning_rate": 1.2218785046728971e-05,
"loss": 0.2069,
"step": 1300
},
{
"epoch": 0.19,
"grad_norm": 5.246530930743759,
"learning_rate": 1.2403831775700934e-05,
"loss": 0.23,
"step": 1320
},
{
"epoch": 0.19,
"grad_norm": 4.212700469456775,
"learning_rate": 1.2588878504672896e-05,
"loss": 0.215,
"step": 1340
},
{
"epoch": 0.19,
"grad_norm": 3.901172847944394,
"learning_rate": 1.277392523364486e-05,
"loss": 0.2159,
"step": 1360
},
{
"epoch": 0.19,
"grad_norm": 3.863116963811273,
"learning_rate": 1.2958971962616822e-05,
"loss": 0.2187,
"step": 1380
},
{
"epoch": 0.2,
"grad_norm": 5.173623334058052,
"learning_rate": 1.3144018691588785e-05,
"loss": 0.2327,
"step": 1400
},
{
"epoch": 0.2,
"grad_norm": 3.4545670342113586,
"learning_rate": 1.3329065420560747e-05,
"loss": 0.2197,
"step": 1420
},
{
"epoch": 0.2,
"grad_norm": 12.081302516012023,
"learning_rate": 1.351411214953271e-05,
"loss": 0.2197,
"step": 1440
},
{
"epoch": 0.2,
"grad_norm": 3.515705158591374,
"learning_rate": 1.3699158878504673e-05,
"loss": 0.2135,
"step": 1460
},
{
"epoch": 0.21,
"grad_norm": 3.5625103063601604,
"learning_rate": 1.3884205607476634e-05,
"loss": 0.2106,
"step": 1480
},
{
"epoch": 0.21,
"grad_norm": 2.8166379370343404,
"learning_rate": 1.4069252336448598e-05,
"loss": 0.2022,
"step": 1500
},
{
"epoch": 0.21,
"grad_norm": 3.319250749357357,
"learning_rate": 1.425429906542056e-05,
"loss": 0.2215,
"step": 1520
},
{
"epoch": 0.22,
"grad_norm": 3.8438255488716826,
"learning_rate": 1.4439345794392523e-05,
"loss": 0.2029,
"step": 1540
},
{
"epoch": 0.22,
"grad_norm": 6.874012546717648,
"learning_rate": 1.4624392523364485e-05,
"loss": 0.2127,
"step": 1560
},
{
"epoch": 0.22,
"grad_norm": 3.340860134089255,
"learning_rate": 1.4809439252336449e-05,
"loss": 0.2111,
"step": 1580
},
{
"epoch": 0.22,
"grad_norm": 4.9714764486240535,
"learning_rate": 1.499448598130841e-05,
"loss": 0.1929,
"step": 1600
},
{
"epoch": 0.23,
"grad_norm": 5.3322988142574355,
"learning_rate": 1.5179532710280374e-05,
"loss": 0.2115,
"step": 1620
},
{
"epoch": 0.23,
"grad_norm": 3.3819070004986793,
"learning_rate": 1.5364579439252338e-05,
"loss": 0.2045,
"step": 1640
},
{
"epoch": 0.23,
"grad_norm": 3.5044078319053185,
"learning_rate": 1.55496261682243e-05,
"loss": 0.1939,
"step": 1660
},
{
"epoch": 0.24,
"grad_norm": 3.608333024398638,
"learning_rate": 1.573467289719626e-05,
"loss": 0.1879,
"step": 1680
},
{
"epoch": 0.24,
"grad_norm": 3.289870210539812,
"learning_rate": 1.5919719626168223e-05,
"loss": 0.1961,
"step": 1700
},
{
"epoch": 0.24,
"grad_norm": 2.45841332213879,
"learning_rate": 1.6104766355140185e-05,
"loss": 0.197,
"step": 1720
},
{
"epoch": 0.24,
"grad_norm": 3.5031262468374926,
"learning_rate": 1.628981308411215e-05,
"loss": 0.1917,
"step": 1740
},
{
"epoch": 0.25,
"grad_norm": 4.246786585681142,
"learning_rate": 1.6474859813084112e-05,
"loss": 0.2047,
"step": 1760
},
{
"epoch": 0.25,
"grad_norm": 3.250792040047206,
"learning_rate": 1.6659906542056074e-05,
"loss": 0.2029,
"step": 1780
},
{
"epoch": 0.25,
"grad_norm": 3.32308980237606,
"learning_rate": 1.6844953271028036e-05,
"loss": 0.2036,
"step": 1800
},
{
"epoch": 0.26,
"grad_norm": 3.6658799599048697,
"learning_rate": 1.703e-05,
"loss": 0.19,
"step": 1820
},
{
"epoch": 0.26,
"grad_norm": 4.176278180207415,
"learning_rate": 1.7215046728971963e-05,
"loss": 0.2032,
"step": 1840
},
{
"epoch": 0.26,
"grad_norm": 3.367015435940533,
"learning_rate": 1.740009345794393e-05,
"loss": 0.2022,
"step": 1860
},
{
"epoch": 0.26,
"grad_norm": 4.672794315382092,
"learning_rate": 1.758514018691589e-05,
"loss": 0.1919,
"step": 1880
},
{
"epoch": 0.27,
"grad_norm": 2.9512188931045236,
"learning_rate": 1.7770186915887852e-05,
"loss": 0.2034,
"step": 1900
},
{
"epoch": 0.27,
"grad_norm": 4.5492480400723965,
"learning_rate": 1.7955233644859814e-05,
"loss": 0.2027,
"step": 1920
},
{
"epoch": 0.27,
"grad_norm": 18.40600129819246,
"learning_rate": 1.814028037383178e-05,
"loss": 0.2237,
"step": 1940
},
{
"epoch": 0.27,
"grad_norm": 4.483521064505469,
"learning_rate": 1.832532710280374e-05,
"loss": 0.206,
"step": 1960
},
{
"epoch": 0.28,
"grad_norm": 3.1092566079454964,
"learning_rate": 1.8510373831775703e-05,
"loss": 0.1977,
"step": 1980
},
{
"epoch": 0.28,
"grad_norm": 3.3220600869694614,
"learning_rate": 1.8695420560747665e-05,
"loss": 0.2063,
"step": 2000
},
{
"epoch": 0.28,
"grad_norm": 3.8907346901502016,
"learning_rate": 1.888046728971963e-05,
"loss": 0.1977,
"step": 2020
},
{
"epoch": 0.29,
"grad_norm": 3.662348169297832,
"learning_rate": 1.9065514018691592e-05,
"loss": 0.1987,
"step": 2040
},
{
"epoch": 0.29,
"grad_norm": 4.059481864151026,
"learning_rate": 1.9250560747663554e-05,
"loss": 0.2048,
"step": 2060
},
{
"epoch": 0.29,
"grad_norm": 4.636029263710959,
"learning_rate": 1.9435607476635516e-05,
"loss": 0.2035,
"step": 2080
},
{
"epoch": 0.29,
"grad_norm": 3.5136376761467996,
"learning_rate": 1.9620654205607477e-05,
"loss": 0.2075,
"step": 2100
},
{
"epoch": 0.3,
"grad_norm": 4.195609140830694,
"learning_rate": 1.9805700934579443e-05,
"loss": 0.1972,
"step": 2120
},
{
"epoch": 0.3,
"grad_norm": 2.9822583106855167,
"learning_rate": 1.9990747663551405e-05,
"loss": 0.1865,
"step": 2140
},
{
"epoch": 0.3,
"grad_norm": 3.7335704750856973,
"learning_rate": 1.9999988060278414e-05,
"loss": 0.197,
"step": 2160
},
{
"epoch": 0.31,
"grad_norm": 2.8405980678578366,
"learning_rate": 1.9999952241142166e-05,
"loss": 0.2068,
"step": 2180
},
{
"epoch": 0.31,
"grad_norm": 3.8657576566737646,
"learning_rate": 1.9999892542676804e-05,
"loss": 0.2049,
"step": 2200
},
{
"epoch": 0.31,
"grad_norm": 4.186876620679429,
"learning_rate": 1.9999808965024894e-05,
"loss": 0.2081,
"step": 2220
},
{
"epoch": 0.31,
"grad_norm": 5.72774777499626,
"learning_rate": 1.9999701508386035e-05,
"loss": 0.2086,
"step": 2240
},
{
"epoch": 0.32,
"grad_norm": 5.751340626875412,
"learning_rate": 1.9999570173016853e-05,
"loss": 0.2006,
"step": 2260
},
{
"epoch": 0.32,
"grad_norm": 3.208700547114019,
"learning_rate": 1.9999414959231005e-05,
"loss": 0.2008,
"step": 2280
},
{
"epoch": 0.32,
"grad_norm": 3.113010447366047,
"learning_rate": 1.999923586739917e-05,
"loss": 0.1908,
"step": 2300
},
{
"epoch": 0.33,
"grad_norm": 4.099631707002062,
"learning_rate": 1.9999032897949043e-05,
"loss": 0.1938,
"step": 2320
},
{
"epoch": 0.33,
"grad_norm": 3.543357953918798,
"learning_rate": 1.9998806051365364e-05,
"loss": 0.1925,
"step": 2340
},
{
"epoch": 0.33,
"grad_norm": 4.294951185097806,
"learning_rate": 1.9998555328189877e-05,
"loss": 0.1917,
"step": 2360
},
{
"epoch": 0.33,
"grad_norm": 2.8667409273627604,
"learning_rate": 1.9998280729021356e-05,
"loss": 0.1894,
"step": 2380
},
{
"epoch": 0.34,
"grad_norm": 3.7775587239677564,
"learning_rate": 1.9997982254515594e-05,
"loss": 0.2013,
"step": 2400
},
{
"epoch": 0.34,
"grad_norm": 3.170937385472613,
"learning_rate": 1.9997659905385407e-05,
"loss": 0.1856,
"step": 2420
},
{
"epoch": 0.34,
"grad_norm": 3.2283581777813706,
"learning_rate": 1.9997313682400616e-05,
"loss": 0.1975,
"step": 2440
},
{
"epoch": 0.34,
"grad_norm": 3.2855714950573365,
"learning_rate": 1.999694358638807e-05,
"loss": 0.1848,
"step": 2460
},
{
"epoch": 0.35,
"grad_norm": 3.863058826303762,
"learning_rate": 1.999654961823163e-05,
"loss": 0.1962,
"step": 2480
},
{
"epoch": 0.35,
"grad_norm": 6.122863085359703,
"learning_rate": 1.9996131778872157e-05,
"loss": 0.1826,
"step": 2500
},
{
"epoch": 0.35,
"grad_norm": 7.479137650825796,
"learning_rate": 1.999569006930753e-05,
"loss": 0.2059,
"step": 2520
},
{
"epoch": 0.36,
"grad_norm": 3.3632282248469587,
"learning_rate": 1.999522449059263e-05,
"loss": 0.1766,
"step": 2540
},
{
"epoch": 0.36,
"grad_norm": 3.5776678734504865,
"learning_rate": 1.999473504383935e-05,
"loss": 0.2058,
"step": 2560
},
{
"epoch": 0.36,
"grad_norm": 3.7860849538340955,
"learning_rate": 1.9994221730216573e-05,
"loss": 0.201,
"step": 2580
},
{
"epoch": 0.36,
"grad_norm": 4.81880844615121,
"learning_rate": 1.999368455095019e-05,
"loss": 0.2117,
"step": 2600
},
{
"epoch": 0.37,
"grad_norm": 2.9215084396780973,
"learning_rate": 1.9993123507323082e-05,
"loss": 0.206,
"step": 2620
},
{
"epoch": 0.37,
"grad_norm": 3.9148674477929726,
"learning_rate": 1.999253860067512e-05,
"loss": 0.2024,
"step": 2640
},
{
"epoch": 0.37,
"grad_norm": 3.2034807705433805,
"learning_rate": 1.9991929832403177e-05,
"loss": 0.1995,
"step": 2660
},
{
"epoch": 0.38,
"grad_norm": 3.339765795416994,
"learning_rate": 1.9991297203961097e-05,
"loss": 0.1955,
"step": 2680
},
{
"epoch": 0.38,
"grad_norm": 3.577891512919677,
"learning_rate": 1.9990640716859712e-05,
"loss": 0.1989,
"step": 2700
},
{
"epoch": 0.38,
"grad_norm": 2.8960487413677845,
"learning_rate": 1.9989960372666837e-05,
"loss": 0.1893,
"step": 2720
},
{
"epoch": 0.38,
"grad_norm": 2.2620031411902137,
"learning_rate": 1.9989256173007253e-05,
"loss": 0.1819,
"step": 2740
},
{
"epoch": 0.39,
"grad_norm": 3.5388703620408384,
"learning_rate": 1.9988528119562726e-05,
"loss": 0.2136,
"step": 2760
},
{
"epoch": 0.39,
"grad_norm": 2.9521351926596697,
"learning_rate": 1.998777621407197e-05,
"loss": 0.1949,
"step": 2780
},
{
"epoch": 0.39,
"grad_norm": 2.4898752230284997,
"learning_rate": 1.9987000458330687e-05,
"loss": 0.1819,
"step": 2800
},
{
"epoch": 0.4,
"grad_norm": 3.3050431585451636,
"learning_rate": 1.998620085419151e-05,
"loss": 0.2115,
"step": 2820
},
{
"epoch": 0.4,
"grad_norm": 2.675538613115082,
"learning_rate": 1.9985377403564047e-05,
"loss": 0.1772,
"step": 2840
},
{
"epoch": 0.4,
"grad_norm": 3.3229381277573196,
"learning_rate": 1.998453010841485e-05,
"loss": 0.2242,
"step": 2860
},
{
"epoch": 0.4,
"grad_norm": 3.180651019652027,
"learning_rate": 1.998365897076741e-05,
"loss": 0.1806,
"step": 2880
},
{
"epoch": 0.41,
"grad_norm": 4.247821249396129,
"learning_rate": 1.9982763992702167e-05,
"loss": 0.1755,
"step": 2900
},
{
"epoch": 0.41,
"grad_norm": 3.8883580723539763,
"learning_rate": 1.9981845176356497e-05,
"loss": 0.2,
"step": 2920
},
{
"epoch": 0.41,
"grad_norm": 3.0966656749820936,
"learning_rate": 1.998090252392469e-05,
"loss": 0.2007,
"step": 2940
},
{
"epoch": 0.41,
"grad_norm": 9.083903956790351,
"learning_rate": 1.9979936037657984e-05,
"loss": 0.1886,
"step": 2960
},
{
"epoch": 0.42,
"grad_norm": 3.599987422073432,
"learning_rate": 1.997894571986452e-05,
"loss": 0.1774,
"step": 2980
},
{
"epoch": 0.42,
"grad_norm": 3.105323029943953,
"learning_rate": 1.9977931572909357e-05,
"loss": 0.1896,
"step": 3000
},
{
"epoch": 0.42,
"grad_norm": 3.854148754731217,
"learning_rate": 1.997689359921446e-05,
"loss": 0.2011,
"step": 3020
},
{
"epoch": 0.43,
"grad_norm": 3.1425993677420627,
"learning_rate": 1.997583180125871e-05,
"loss": 0.1798,
"step": 3040
},
{
"epoch": 0.43,
"grad_norm": 4.715875015052114,
"learning_rate": 1.9974746181577876e-05,
"loss": 0.1789,
"step": 3060
},
{
"epoch": 0.43,
"grad_norm": 3.278262713133406,
"learning_rate": 1.9973636742764604e-05,
"loss": 0.1934,
"step": 3080
},
{
"epoch": 0.43,
"grad_norm": 4.634152609388742,
"learning_rate": 1.9972503487468448e-05,
"loss": 0.1779,
"step": 3100
},
{
"epoch": 0.44,
"grad_norm": 2.9495097642016876,
"learning_rate": 1.9971346418395823e-05,
"loss": 0.1995,
"step": 3120
},
{
"epoch": 0.44,
"grad_norm": 3.8346391333501018,
"learning_rate": 1.9970165538310022e-05,
"loss": 0.1742,
"step": 3140
},
{
"epoch": 0.44,
"grad_norm": 0.2372528380590026,
"learning_rate": 1.996896085003121e-05,
"loss": 0.1715,
"step": 3160
},
{
"epoch": 0.45,
"grad_norm": 2.3982539898876865,
"learning_rate": 1.9967732356436397e-05,
"loss": 0.1824,
"step": 3180
},
{
"epoch": 0.45,
"grad_norm": 3.021270211735749,
"learning_rate": 1.996648006045945e-05,
"loss": 0.1724,
"step": 3200
},
{
"epoch": 0.45,
"grad_norm": 3.218541555229141,
"learning_rate": 1.9965203965091086e-05,
"loss": 0.1895,
"step": 3220
},
{
"epoch": 0.45,
"grad_norm": 2.028124943573044,
"learning_rate": 1.9963904073378852e-05,
"loss": 0.1711,
"step": 3240
},
{
"epoch": 0.46,
"grad_norm": 3.1891582887684193,
"learning_rate": 1.996258038842712e-05,
"loss": 0.1856,
"step": 3260
},
{
"epoch": 0.46,
"grad_norm": 3.884703147082746,
"learning_rate": 1.9961232913397107e-05,
"loss": 0.1874,
"step": 3280
},
{
"epoch": 0.46,
"grad_norm": 2.445509345081033,
"learning_rate": 1.995986165150682e-05,
"loss": 0.1974,
"step": 3300
},
{
"epoch": 0.47,
"grad_norm": 2.376178449088288,
"learning_rate": 1.9958466606031087e-05,
"loss": 0.1864,
"step": 3320
},
{
"epoch": 0.47,
"grad_norm": 2.3656935809133044,
"learning_rate": 1.9957047780301527e-05,
"loss": 0.1775,
"step": 3340
},
{
"epoch": 0.47,
"grad_norm": 2.854832109902861,
"learning_rate": 1.9955605177706563e-05,
"loss": 0.1832,
"step": 3360
},
{
"epoch": 0.47,
"grad_norm": 3.747222762079925,
"learning_rate": 1.995413880169139e-05,
"loss": 0.1766,
"step": 3380
},
{
"epoch": 0.48,
"grad_norm": 3.4619845836434995,
"learning_rate": 1.9952648655757984e-05,
"loss": 0.1992,
"step": 3400
},
{
"epoch": 0.48,
"grad_norm": 8.299795897398717,
"learning_rate": 1.9951134743465083e-05,
"loss": 0.1868,
"step": 3420
},
{
"epoch": 0.48,
"grad_norm": 3.989784069671869,
"learning_rate": 1.994959706842819e-05,
"loss": 0.1972,
"step": 3440
},
{
"epoch": 0.49,
"grad_norm": 2.442836495123392,
"learning_rate": 1.9948035634319554e-05,
"loss": 0.1879,
"step": 3460
},
{
"epoch": 0.49,
"grad_norm": 2.8857456575169493,
"learning_rate": 1.9946450444868168e-05,
"loss": 0.177,
"step": 3480
},
{
"epoch": 0.49,
"grad_norm": 3.1516856814866063,
"learning_rate": 1.9944841503859752e-05,
"loss": 0.1764,
"step": 3500
},
{
"epoch": 0.49,
"grad_norm": 3.648426521497133,
"learning_rate": 1.994320881513675e-05,
"loss": 0.1951,
"step": 3520
},
{
"epoch": 0.5,
"grad_norm": 3.119631007945078,
"learning_rate": 1.9941552382598326e-05,
"loss": 0.1879,
"step": 3540
},
{
"epoch": 0.5,
"grad_norm": 2.601513836379742,
"learning_rate": 1.993987221020034e-05,
"loss": 0.185,
"step": 3560
},
{
"epoch": 0.5,
"grad_norm": 3.6061574623227988,
"learning_rate": 1.9938168301955354e-05,
"loss": 0.1915,
"step": 3580
},
{
"epoch": 0.5,
"grad_norm": 3.3502344020995354,
"learning_rate": 1.9936440661932613e-05,
"loss": 0.195,
"step": 3600
},
{
"epoch": 0.51,
"grad_norm": 3.2199091629578924,
"learning_rate": 1.9934689294258037e-05,
"loss": 0.1923,
"step": 3620
},
{
"epoch": 0.51,
"grad_norm": 3.525187596949837,
"learning_rate": 1.993291420311421e-05,
"loss": 0.1921,
"step": 3640
},
{
"epoch": 0.51,
"grad_norm": 3.96466671402784,
"learning_rate": 1.9931115392740383e-05,
"loss": 0.1974,
"step": 3660
},
{
"epoch": 0.52,
"grad_norm": 3.4114491392210713,
"learning_rate": 1.9929292867432435e-05,
"loss": 0.1814,
"step": 3680
},
{
"epoch": 0.52,
"grad_norm": 2.8956428483478396,
"learning_rate": 1.9927446631542894e-05,
"loss": 0.1907,
"step": 3700
},
{
"epoch": 0.52,
"grad_norm": 5.004579163807035,
"learning_rate": 1.9925576689480915e-05,
"loss": 0.1773,
"step": 3720
},
{
"epoch": 0.52,
"grad_norm": 2.645550259643549,
"learning_rate": 1.992368304571225e-05,
"loss": 0.1824,
"step": 3740
},
{
"epoch": 0.53,
"grad_norm": 2.6946320932684875,
"learning_rate": 1.992176570475928e-05,
"loss": 0.2026,
"step": 3760
},
{
"epoch": 0.53,
"grad_norm": 3.5657388787635416,
"learning_rate": 1.991982467120096e-05,
"loss": 0.1941,
"step": 3780
},
{
"epoch": 0.53,
"grad_norm": 4.334836967948886,
"learning_rate": 1.9917859949672828e-05,
"loss": 0.191,
"step": 3800
},
{
"epoch": 0.54,
"grad_norm": 3.41391640357102,
"learning_rate": 1.9915871544867012e-05,
"loss": 0.1919,
"step": 3820
},
{
"epoch": 0.54,
"grad_norm": 3.1085888045991386,
"learning_rate": 1.9913859461532177e-05,
"loss": 0.1825,
"step": 3840
},
{
"epoch": 0.54,
"grad_norm": 3.6245604939232132,
"learning_rate": 1.9911823704473546e-05,
"loss": 0.1778,
"step": 3860
},
{
"epoch": 0.54,
"grad_norm": 4.041287722306811,
"learning_rate": 1.990976427855289e-05,
"loss": 0.1873,
"step": 3880
},
{
"epoch": 0.55,
"grad_norm": 3.1845423121948278,
"learning_rate": 1.9907681188688482e-05,
"loss": 0.1835,
"step": 3900
},
{
"epoch": 0.55,
"grad_norm": 4.132646121045449,
"learning_rate": 1.990557443985513e-05,
"loss": 0.2008,
"step": 3920
},
{
"epoch": 0.55,
"grad_norm": 3.4519916590548956,
"learning_rate": 1.9903444037084133e-05,
"loss": 0.1899,
"step": 3940
},
{
"epoch": 0.56,
"grad_norm": 2.653297309087807,
"learning_rate": 1.9901289985463283e-05,
"loss": 0.1806,
"step": 3960
},
{
"epoch": 0.56,
"grad_norm": 3.307457738997048,
"learning_rate": 1.9899112290136855e-05,
"loss": 0.1887,
"step": 3980
},
{
"epoch": 0.56,
"grad_norm": 3.6268024567352253,
"learning_rate": 1.9896910956305574e-05,
"loss": 0.1787,
"step": 4000
},
{
"epoch": 0.56,
"grad_norm": 3.331815294465332,
"learning_rate": 1.989468598922664e-05,
"loss": 0.1741,
"step": 4020
},
{
"epoch": 0.57,
"grad_norm": 4.187774421468107,
"learning_rate": 1.9892437394213678e-05,
"loss": 0.1983,
"step": 4040
},
{
"epoch": 0.57,
"grad_norm": 3.7206646774588332,
"learning_rate": 1.9890165176636736e-05,
"loss": 0.1874,
"step": 4060
},
{
"epoch": 0.57,
"grad_norm": 3.2845949108425105,
"learning_rate": 1.9887869341922296e-05,
"loss": 0.1883,
"step": 4080
},
{
"epoch": 0.57,
"grad_norm": 2.580203703741531,
"learning_rate": 1.988554989555323e-05,
"loss": 0.1873,
"step": 4100
},
{
"epoch": 0.58,
"grad_norm": 3.540337137235391,
"learning_rate": 1.98832068430688e-05,
"loss": 0.167,
"step": 4120
},
{
"epoch": 0.58,
"grad_norm": 5.842048601334258,
"learning_rate": 1.988084019006464e-05,
"loss": 0.1783,
"step": 4140
},
{
"epoch": 0.58,
"grad_norm": 2.5378369939115504,
"learning_rate": 1.9878449942192755e-05,
"loss": 0.1702,
"step": 4160
},
{
"epoch": 0.59,
"grad_norm": 2.701425006780818,
"learning_rate": 1.9876036105161498e-05,
"loss": 0.168,
"step": 4180
},
{
"epoch": 0.59,
"grad_norm": 3.090708923743249,
"learning_rate": 1.9873598684735546e-05,
"loss": 0.169,
"step": 4200
},
{
"epoch": 0.59,
"grad_norm": 4.53332640329467,
"learning_rate": 1.9871137686735907e-05,
"loss": 0.1866,
"step": 4220
},
{
"epoch": 0.59,
"grad_norm": 2.3722892148248733,
"learning_rate": 1.9868653117039902e-05,
"loss": 0.1777,
"step": 4240
},
{
"epoch": 0.6,
"grad_norm": 2.7870109129036074,
"learning_rate": 1.9866144981581132e-05,
"loss": 0.1819,
"step": 4260
},
{
"epoch": 0.6,
"grad_norm": 2.9397160713475645,
"learning_rate": 1.9863613286349486e-05,
"loss": 0.1714,
"step": 4280
},
{
"epoch": 0.6,
"grad_norm": 17.31460811848986,
"learning_rate": 1.9861058037391118e-05,
"loss": 0.1813,
"step": 4300
},
{
"epoch": 0.61,
"grad_norm": 6.045029153934977,
"learning_rate": 1.985847924080842e-05,
"loss": 0.1828,
"step": 4320
},
{
"epoch": 0.61,
"grad_norm": 2.6773267306300865,
"learning_rate": 1.9855876902760046e-05,
"loss": 0.1784,
"step": 4340
},
{
"epoch": 0.61,
"grad_norm": 5.2228019654083235,
"learning_rate": 1.9853251029460844e-05,
"loss": 0.1764,
"step": 4360
},
{
"epoch": 0.61,
"grad_norm": 3.063926208982017,
"learning_rate": 1.9850601627181885e-05,
"loss": 0.178,
"step": 4380
},
{
"epoch": 0.62,
"grad_norm": 3.0456294717331187,
"learning_rate": 1.984792870225043e-05,
"loss": 0.1758,
"step": 4400
},
{
"epoch": 0.62,
"grad_norm": 3.0217671671960664,
"learning_rate": 1.98452322610499e-05,
"loss": 0.1575,
"step": 4420
},
{
"epoch": 0.62,
"grad_norm": 3.2515513312218083,
"learning_rate": 1.984251231001991e-05,
"loss": 0.1758,
"step": 4440
},
{
"epoch": 0.63,
"grad_norm": 2.816509186745791,
"learning_rate": 1.983976885565619e-05,
"loss": 0.1754,
"step": 4460
},
{
"epoch": 0.63,
"grad_norm": 2.879544853697043,
"learning_rate": 1.983700190451061e-05,
"loss": 0.1666,
"step": 4480
},
{
"epoch": 0.63,
"grad_norm": 4.239525621975842,
"learning_rate": 1.983421146319116e-05,
"loss": 0.1638,
"step": 4500
},
{
"epoch": 0.63,
"grad_norm": 3.4223430435381266,
"learning_rate": 1.983139753836192e-05,
"loss": 0.1884,
"step": 4520
},
{
"epoch": 0.64,
"grad_norm": 2.434002249585807,
"learning_rate": 1.9828560136743063e-05,
"loss": 0.1836,
"step": 4540
},
{
"epoch": 0.64,
"grad_norm": 11.273460848143701,
"learning_rate": 1.9825699265110822e-05,
"loss": 0.1828,
"step": 4560
},
{
"epoch": 0.64,
"grad_norm": 2.8411685838962706,
"learning_rate": 1.9822814930297486e-05,
"loss": 0.1722,
"step": 4580
},
{
"epoch": 0.64,
"grad_norm": 15.676474781412614,
"learning_rate": 1.981990713919137e-05,
"loss": 0.1745,
"step": 4600
},
{
"epoch": 0.65,
"grad_norm": 3.9204743456434747,
"learning_rate": 1.981697589873681e-05,
"loss": 0.1649,
"step": 4620
},
{
"epoch": 0.65,
"grad_norm": 2.5584019712835353,
"learning_rate": 1.9814021215934148e-05,
"loss": 0.1667,
"step": 4640
},
{
"epoch": 0.65,
"grad_norm": 2.1414636236143325,
"learning_rate": 1.981104309783971e-05,
"loss": 0.176,
"step": 4660
},
{
"epoch": 0.66,
"grad_norm": 2.1512634557095933,
"learning_rate": 1.980804155156578e-05,
"loss": 0.1705,
"step": 4680
},
{
"epoch": 0.66,
"grad_norm": 2.654510556813465,
"learning_rate": 1.980501658428061e-05,
"loss": 0.1757,
"step": 4700
},
{
"epoch": 0.66,
"grad_norm": 2.9013380975590652,
"learning_rate": 1.980196820320837e-05,
"loss": 0.1759,
"step": 4720
},
{
"epoch": 0.66,
"grad_norm": 2.9545228263682044,
"learning_rate": 1.9798896415629157e-05,
"loss": 0.1783,
"step": 4740
},
{
"epoch": 0.67,
"grad_norm": 3.543349745010547,
"learning_rate": 1.9795801228878952e-05,
"loss": 0.1836,
"step": 4760
},
{
"epoch": 0.67,
"grad_norm": 2.2937476217896817,
"learning_rate": 1.979268265034964e-05,
"loss": 0.1821,
"step": 4780
},
{
"epoch": 0.67,
"grad_norm": 4.062304448042868,
"learning_rate": 1.9789540687488945e-05,
"loss": 0.1597,
"step": 4800
},
{
"epoch": 0.68,
"grad_norm": 3.4206144894020754,
"learning_rate": 1.9786375347800465e-05,
"loss": 0.1831,
"step": 4820
},
{
"epoch": 0.68,
"grad_norm": 3.267427375996239,
"learning_rate": 1.9783186638843598e-05,
"loss": 0.1607,
"step": 4840
},
{
"epoch": 0.68,
"grad_norm": 3.7545232831370434,
"learning_rate": 1.977997456823357e-05,
"loss": 0.1912,
"step": 4860
},
{
"epoch": 0.68,
"grad_norm": 3.6111110459026103,
"learning_rate": 1.9776739143641392e-05,
"loss": 0.1775,
"step": 4880
},
{
"epoch": 0.69,
"grad_norm": 3.2964692792694854,
"learning_rate": 1.9773480372793855e-05,
"loss": 0.1705,
"step": 4900
},
{
"epoch": 0.69,
"grad_norm": 2.500103090068091,
"learning_rate": 1.977019826347349e-05,
"loss": 0.1947,
"step": 4920
},
{
"epoch": 0.69,
"grad_norm": 4.150835913612857,
"learning_rate": 1.9766892823518595e-05,
"loss": 0.1885,
"step": 4940
},
{
"epoch": 0.7,
"grad_norm": 2.4040456806419126,
"learning_rate": 1.9763564060823145e-05,
"loss": 0.1933,
"step": 4960
},
{
"epoch": 0.7,
"grad_norm": 7.116540270121414,
"learning_rate": 1.9760211983336846e-05,
"loss": 0.1669,
"step": 4980
},
{
"epoch": 0.7,
"grad_norm": 3.27540722127951,
"learning_rate": 1.975683659906507e-05,
"loss": 0.18,
"step": 5000
},
{
"epoch": 0.7,
"grad_norm": 15.517488519447078,
"learning_rate": 1.975343791606885e-05,
"loss": 0.1918,
"step": 5020
},
{
"epoch": 0.71,
"grad_norm": 2.897248798109084,
"learning_rate": 1.975001594246487e-05,
"loss": 0.1558,
"step": 5040
},
{
"epoch": 0.71,
"grad_norm": 2.834736551359366,
"learning_rate": 1.9746570686425427e-05,
"loss": 0.1627,
"step": 5060
},
{
"epoch": 0.71,
"grad_norm": 4.173821955789194,
"learning_rate": 1.974310215617842e-05,
"loss": 0.1756,
"step": 5080
},
{
"epoch": 0.71,
"grad_norm": 2.992296660102903,
"learning_rate": 1.973961036000734e-05,
"loss": 0.1807,
"step": 5100
},
{
"epoch": 0.72,
"grad_norm": 3.3528000474169324,
"learning_rate": 1.9736095306251232e-05,
"loss": 0.1874,
"step": 5120
},
{
"epoch": 0.72,
"grad_norm": 3.334344058979202,
"learning_rate": 1.9732557003304686e-05,
"loss": 0.1899,
"step": 5140
},
{
"epoch": 0.72,
"grad_norm": 2.860455585019566,
"learning_rate": 1.9728995459617824e-05,
"loss": 0.1567,
"step": 5160
},
{
"epoch": 0.73,
"grad_norm": 5.958156271840138,
"learning_rate": 1.972541068369626e-05,
"loss": 0.1733,
"step": 5180
},
{
"epoch": 0.73,
"grad_norm": 2.4787961111237973,
"learning_rate": 1.9721802684101095e-05,
"loss": 0.1858,
"step": 5200
},
{
"epoch": 0.73,
"grad_norm": 2.3998338443142604,
"learning_rate": 1.9718171469448896e-05,
"loss": 0.178,
"step": 5220
},
{
"epoch": 0.73,
"grad_norm": 3.197290531124358,
"learning_rate": 1.9714517048411662e-05,
"loss": 0.1638,
"step": 5240
},
{
"epoch": 0.74,
"grad_norm": 2.435817724859357,
"learning_rate": 1.971083942971683e-05,
"loss": 0.166,
"step": 5260
},
{
"epoch": 0.74,
"grad_norm": 2.2596390143874303,
"learning_rate": 1.9707138622147218e-05,
"loss": 0.1759,
"step": 5280
},
{
"epoch": 0.74,
"grad_norm": 3.3606753649209007,
"learning_rate": 1.9703414634541038e-05,
"loss": 0.1639,
"step": 5300
},
{
"epoch": 0.75,
"grad_norm": 3.099118131903914,
"learning_rate": 1.969966747579185e-05,
"loss": 0.1846,
"step": 5320
},
{
"epoch": 0.75,
"grad_norm": 2.5845845996591814,
"learning_rate": 1.9695897154848557e-05,
"loss": 0.1854,
"step": 5340
},
{
"epoch": 0.75,
"grad_norm": 3.2375821784352565,
"learning_rate": 1.9692103680715377e-05,
"loss": 0.1639,
"step": 5360
},
{
"epoch": 0.75,
"grad_norm": 2.9657833334457906,
"learning_rate": 1.968828706245182e-05,
"loss": 0.1651,
"step": 5380
},
{
"epoch": 0.76,
"grad_norm": 2.753672315252415,
"learning_rate": 1.968444730917267e-05,
"loss": 0.1706,
"step": 5400
},
{
"epoch": 0.76,
"grad_norm": 2.721605927474235,
"learning_rate": 1.968058443004796e-05,
"loss": 0.1788,
"step": 5420
},
{
"epoch": 0.76,
"grad_norm": 3.4558253712068345,
"learning_rate": 1.9676698434302956e-05,
"loss": 0.1763,
"step": 5440
},
{
"epoch": 0.77,
"grad_norm": 3.458103095667287,
"learning_rate": 1.967278933121812e-05,
"loss": 0.1685,
"step": 5460
},
{
"epoch": 0.77,
"grad_norm": 2.701537619851815,
"learning_rate": 1.9668857130129116e-05,
"loss": 0.1633,
"step": 5480
},
{
"epoch": 0.77,
"grad_norm": 3.1088759438741542,
"learning_rate": 1.9664901840426752e-05,
"loss": 0.1637,
"step": 5500
},
{
"epoch": 0.77,
"grad_norm": 2.2628272309947888,
"learning_rate": 1.9660923471556984e-05,
"loss": 0.1743,
"step": 5520
},
{
"epoch": 0.78,
"grad_norm": 3.2565461769057316,
"learning_rate": 1.9656922033020897e-05,
"loss": 0.1869,
"step": 5540
},
{
"epoch": 0.78,
"grad_norm": 2.729887326049785,
"learning_rate": 1.9652897534374643e-05,
"loss": 0.1832,
"step": 5560
},
{
"epoch": 0.78,
"grad_norm": 3.4960360559572643,
"learning_rate": 1.964884998522947e-05,
"loss": 0.1698,
"step": 5580
},
{
"epoch": 0.79,
"grad_norm": 1.9936482179306132,
"learning_rate": 1.9644779395251673e-05,
"loss": 0.1667,
"step": 5600
},
{
"epoch": 0.79,
"grad_norm": 2.9652832670456295,
"learning_rate": 1.9640685774162553e-05,
"loss": 0.1976,
"step": 5620
},
{
"epoch": 0.79,
"grad_norm": 3.0252002239957063,
"learning_rate": 1.9636569131738438e-05,
"loss": 0.1587,
"step": 5640
},
{
"epoch": 0.79,
"grad_norm": 3.2950965459684816,
"learning_rate": 1.9632429477810616e-05,
"loss": 0.1698,
"step": 5660
},
{
"epoch": 0.8,
"grad_norm": 2.8160632132240964,
"learning_rate": 1.9628266822265345e-05,
"loss": 0.1587,
"step": 5680
},
{
"epoch": 0.8,
"grad_norm": 2.3142226736205322,
"learning_rate": 1.9624081175043802e-05,
"loss": 0.18,
"step": 5700
},
{
"epoch": 0.8,
"grad_norm": 2.466225600975058,
"learning_rate": 1.961987254614209e-05,
"loss": 0.1715,
"step": 5720
},
{
"epoch": 0.8,
"grad_norm": 2.231814644797845,
"learning_rate": 1.9615640945611175e-05,
"loss": 0.1564,
"step": 5740
},
{
"epoch": 0.81,
"grad_norm": 2.0486620607523975,
"learning_rate": 1.9611386383556893e-05,
"loss": 0.1654,
"step": 5760
},
{
"epoch": 0.81,
"grad_norm": 3.336933925213419,
"learning_rate": 1.9607108870139927e-05,
"loss": 0.1629,
"step": 5780
},
{
"epoch": 0.81,
"grad_norm": 2.6663360291384848,
"learning_rate": 1.9602808415575757e-05,
"loss": 0.166,
"step": 5800
},
{
"epoch": 0.82,
"grad_norm": 3.2043231491021738,
"learning_rate": 1.9598485030134652e-05,
"loss": 0.1723,
"step": 5820
},
{
"epoch": 0.82,
"grad_norm": 3.4363565103586358,
"learning_rate": 1.9594138724141657e-05,
"loss": 0.169,
"step": 5840
},
{
"epoch": 0.82,
"grad_norm": 4.047561881786324,
"learning_rate": 1.9589769507976543e-05,
"loss": 0.1884,
"step": 5860
},
{
"epoch": 0.82,
"grad_norm": 2.586658665343163,
"learning_rate": 1.9585377392073795e-05,
"loss": 0.159,
"step": 5880
},
{
"epoch": 0.83,
"grad_norm": 2.8700917681866027,
"learning_rate": 1.958096238692259e-05,
"loss": 0.1589,
"step": 5900
},
{
"epoch": 0.83,
"grad_norm": 3.0157418500171436,
"learning_rate": 1.957652450306678e-05,
"loss": 0.169,
"step": 5920
},
{
"epoch": 0.83,
"grad_norm": 2.624287008912185,
"learning_rate": 1.957206375110483e-05,
"loss": 0.1714,
"step": 5940
},
{
"epoch": 0.84,
"grad_norm": 2.6907468300772104,
"learning_rate": 1.956758014168984e-05,
"loss": 0.173,
"step": 5960
},
{
"epoch": 0.84,
"grad_norm": 3.0837109995997896,
"learning_rate": 1.956307368552949e-05,
"loss": 0.1751,
"step": 5980
},
{
"epoch": 0.84,
"grad_norm": 2.466280124207713,
"learning_rate": 1.955854439338602e-05,
"loss": 0.1692,
"step": 6000
},
{
"epoch": 0.84,
"grad_norm": 2.6498950109737085,
"learning_rate": 1.955399227607621e-05,
"loss": 0.1825,
"step": 6020
},
{
"epoch": 0.85,
"grad_norm": 2.478228757593491,
"learning_rate": 1.954941734447135e-05,
"loss": 0.164,
"step": 6040
},
{
"epoch": 0.85,
"grad_norm": 2.6482682525659724,
"learning_rate": 1.954481960949722e-05,
"loss": 0.1647,
"step": 6060
},
{
"epoch": 0.85,
"grad_norm": 2.2678907969051356,
"learning_rate": 1.9540199082134044e-05,
"loss": 0.1643,
"step": 6080
},
{
"epoch": 0.86,
"grad_norm": 12.195551477078768,
"learning_rate": 1.9535555773416494e-05,
"loss": 0.1785,
"step": 6100
},
{
"epoch": 0.86,
"grad_norm": 2.9127326774157374,
"learning_rate": 1.9530889694433636e-05,
"loss": 0.1745,
"step": 6120
},
{
"epoch": 0.86,
"grad_norm": 2.176968317476819,
"learning_rate": 1.952620085632893e-05,
"loss": 0.1586,
"step": 6140
},
{
"epoch": 0.86,
"grad_norm": 3.022412818932396,
"learning_rate": 1.952148927030017e-05,
"loss": 0.1704,
"step": 6160
},
{
"epoch": 0.87,
"grad_norm": 3.59711371855089,
"learning_rate": 1.951675494759949e-05,
"loss": 0.1594,
"step": 6180
},
{
"epoch": 0.87,
"grad_norm": 2.140677235655155,
"learning_rate": 1.951199789953332e-05,
"loss": 0.1568,
"step": 6200
},
{
"epoch": 0.87,
"grad_norm": 2.202236131499489,
"learning_rate": 1.9507218137462366e-05,
"loss": 0.1737,
"step": 6220
},
{
"epoch": 0.87,
"grad_norm": 2.4482766610431894,
"learning_rate": 1.950241567280157e-05,
"loss": 0.1617,
"step": 6240
},
{
"epoch": 0.88,
"grad_norm": 2.215775335938973,
"learning_rate": 1.94975905170201e-05,
"loss": 0.1672,
"step": 6260
},
{
"epoch": 0.88,
"grad_norm": 2.9637923757547577,
"learning_rate": 1.949274268164131e-05,
"loss": 0.1669,
"step": 6280
},
{
"epoch": 0.88,
"grad_norm": 2.631304572257579,
"learning_rate": 1.9487872178242715e-05,
"loss": 0.1668,
"step": 6300
},
{
"epoch": 0.89,
"grad_norm": 3.844536925607983,
"learning_rate": 1.948297901845598e-05,
"loss": 0.1677,
"step": 6320
},
{
"epoch": 0.89,
"grad_norm": 2.227377866057559,
"learning_rate": 1.9478063213966853e-05,
"loss": 0.1708,
"step": 6340
},
{
"epoch": 0.89,
"grad_norm": 2.841537621067319,
"learning_rate": 1.9473124776515185e-05,
"loss": 0.1693,
"step": 6360
},
{
"epoch": 0.89,
"grad_norm": 2.066597307743588,
"learning_rate": 1.9468163717894866e-05,
"loss": 0.1701,
"step": 6380
},
{
"epoch": 0.9,
"grad_norm": 2.7031979413196576,
"learning_rate": 1.9463180049953814e-05,
"loss": 0.1677,
"step": 6400
},
{
"epoch": 0.9,
"grad_norm": 2.6315076180879995,
"learning_rate": 1.9458173784593936e-05,
"loss": 0.1691,
"step": 6420
},
{
"epoch": 0.9,
"grad_norm": 2.4342973650206408,
"learning_rate": 1.945314493377112e-05,
"loss": 0.16,
"step": 6440
},
{
"epoch": 0.91,
"grad_norm": 3.261599016561727,
"learning_rate": 1.944809350949517e-05,
"loss": 0.1693,
"step": 6460
},
{
"epoch": 0.91,
"grad_norm": 1.9158677288063537,
"learning_rate": 1.9443019523829825e-05,
"loss": 0.178,
"step": 6480
},
{
"epoch": 0.91,
"grad_norm": 2.889106805440004,
"learning_rate": 1.9437922988892683e-05,
"loss": 0.1668,
"step": 6500
},
{
"epoch": 0.91,
"grad_norm": 2.0931458242023493,
"learning_rate": 1.9432803916855204e-05,
"loss": 0.1699,
"step": 6520
},
{
"epoch": 0.92,
"grad_norm": 2.2367781939986693,
"learning_rate": 1.9427662319942674e-05,
"loss": 0.1489,
"step": 6540
},
{
"epoch": 0.92,
"grad_norm": 2.7162488635605437,
"learning_rate": 1.9422498210434165e-05,
"loss": 0.1774,
"step": 6560
},
{
"epoch": 0.92,
"grad_norm": 2.810866750576386,
"learning_rate": 1.9417311600662516e-05,
"loss": 0.1594,
"step": 6580
},
{
"epoch": 0.93,
"grad_norm": 2.689944020933223,
"learning_rate": 1.94121025030143e-05,
"loss": 0.1713,
"step": 6600
},
{
"epoch": 0.93,
"grad_norm": 2.5953451559916356,
"learning_rate": 1.94068709299298e-05,
"loss": 0.1642,
"step": 6620
},
{
"epoch": 0.93,
"grad_norm": 2.579493026750592,
"learning_rate": 1.9401616893902968e-05,
"loss": 0.1652,
"step": 6640
},
{
"epoch": 0.93,
"grad_norm": 2.7325791047172996,
"learning_rate": 1.9396340407481403e-05,
"loss": 0.1501,
"step": 6660
},
{
"epoch": 0.94,
"grad_norm": 3.5718650196967277,
"learning_rate": 1.9391041483266323e-05,
"loss": 0.1689,
"step": 6680
},
{
"epoch": 0.94,
"grad_norm": 2.7282794257609533,
"learning_rate": 1.938572013391253e-05,
"loss": 0.1603,
"step": 6700
},
{
"epoch": 0.94,
"grad_norm": 3.0463500524345153,
"learning_rate": 1.9380376372128377e-05,
"loss": 0.1472,
"step": 6720
},
{
"epoch": 0.94,
"grad_norm": 2.370280948195058,
"learning_rate": 1.937501021067575e-05,
"loss": 0.1655,
"step": 6740
},
{
"epoch": 0.95,
"grad_norm": 2.0522339534831553,
"learning_rate": 1.9369621662370023e-05,
"loss": 0.1578,
"step": 6760
},
{
"epoch": 0.95,
"grad_norm": 3.0773650649708473,
"learning_rate": 1.936421074008004e-05,
"loss": 0.1714,
"step": 6780
},
{
"epoch": 0.95,
"grad_norm": 3.9660532041550423,
"learning_rate": 1.9358777456728072e-05,
"loss": 0.1639,
"step": 6800
},
{
"epoch": 0.96,
"grad_norm": 2.9238454121274184,
"learning_rate": 1.935332182528979e-05,
"loss": 0.1696,
"step": 6820
},
{
"epoch": 0.96,
"grad_norm": 2.768302957167946,
"learning_rate": 1.9347843858794243e-05,
"loss": 0.1571,
"step": 6840
},
{
"epoch": 0.96,
"grad_norm": 1.5457418125064697,
"learning_rate": 1.9342343570323825e-05,
"loss": 0.1479,
"step": 6860
},
{
"epoch": 0.96,
"grad_norm": 2.0257799883396093,
"learning_rate": 1.9336820973014227e-05,
"loss": 0.1643,
"step": 6880
},
{
"epoch": 0.97,
"grad_norm": 2.0318673002812337,
"learning_rate": 1.9331276080054422e-05,
"loss": 0.1489,
"step": 6900
},
{
"epoch": 0.97,
"grad_norm": 2.726328142909849,
"learning_rate": 1.932570890468663e-05,
"loss": 0.1594,
"step": 6920
},
{
"epoch": 0.97,
"grad_norm": 2.990966228920672,
"learning_rate": 1.9320119460206285e-05,
"loss": 0.166,
"step": 6940
},
{
"epoch": 0.98,
"grad_norm": 4.345453820096112,
"learning_rate": 1.9314507759962008e-05,
"loss": 0.1542,
"step": 6960
},
{
"epoch": 0.98,
"grad_norm": 3.1488677457364194,
"learning_rate": 1.9308873817355563e-05,
"loss": 0.1713,
"step": 6980
},
{
"epoch": 0.98,
"grad_norm": 3.219538916042044,
"learning_rate": 1.9303217645841838e-05,
"loss": 0.1492,
"step": 7000
},
{
"epoch": 0.98,
"grad_norm": 3.0131745339076446,
"learning_rate": 1.9297539258928807e-05,
"loss": 0.1612,
"step": 7020
},
{
"epoch": 0.99,
"grad_norm": 2.2164410679488933,
"learning_rate": 1.9291838670177497e-05,
"loss": 0.1706,
"step": 7040
},
{
"epoch": 0.99,
"grad_norm": 2.242976033389265,
"learning_rate": 1.9286115893201955e-05,
"loss": 0.1536,
"step": 7060
},
{
"epoch": 0.99,
"grad_norm": 2.610990560963815,
"learning_rate": 1.928037094166923e-05,
"loss": 0.1676,
"step": 7080
},
{
"epoch": 1.0,
"grad_norm": 2.529344760043821,
"learning_rate": 1.9274603829299304e-05,
"loss": 0.1579,
"step": 7100
},
{
"epoch": 1.0,
"grad_norm": 3.345674905474248,
"learning_rate": 1.926881456986511e-05,
"loss": 0.1676,
"step": 7120
},
{
"epoch": 1.0,
"grad_norm": 2.8957948384825247,
"learning_rate": 1.9263003177192455e-05,
"loss": 0.148,
"step": 7140
},
{
"epoch": 1.0,
"grad_norm": 2.980092916128174,
"learning_rate": 1.925716966516001e-05,
"loss": 0.1407,
"step": 7160
},
{
"epoch": 1.01,
"grad_norm": 2.2603755777260495,
"learning_rate": 1.925131404769927e-05,
"loss": 0.1325,
"step": 7180
},
{
"epoch": 1.01,
"grad_norm": 5.813992216469168,
"learning_rate": 1.9245436338794516e-05,
"loss": 0.1392,
"step": 7200
},
{
"epoch": 1.01,
"grad_norm": 2.520351069534241,
"learning_rate": 1.9239536552482803e-05,
"loss": 0.1453,
"step": 7220
},
{
"epoch": 1.02,
"grad_norm": 2.566425201485392,
"learning_rate": 1.9233614702853892e-05,
"loss": 0.1383,
"step": 7240
},
{
"epoch": 1.02,
"grad_norm": 2.203551555490054,
"learning_rate": 1.9227670804050254e-05,
"loss": 0.1374,
"step": 7260
},
{
"epoch": 1.02,
"grad_norm": 2.623737213118146,
"learning_rate": 1.9221704870267e-05,
"loss": 0.1291,
"step": 7280
},
{
"epoch": 1.02,
"grad_norm": 2.3896072315549794,
"learning_rate": 1.9215716915751877e-05,
"loss": 0.1377,
"step": 7300
},
{
"epoch": 1.03,
"grad_norm": 2.2491299748499,
"learning_rate": 1.9209706954805215e-05,
"loss": 0.1272,
"step": 7320
},
{
"epoch": 1.03,
"grad_norm": 2.607481857924288,
"learning_rate": 1.92036750017799e-05,
"loss": 0.1459,
"step": 7340
},
{
"epoch": 1.03,
"grad_norm": 4.047819874159741,
"learning_rate": 1.9197621071081346e-05,
"loss": 0.145,
"step": 7360
},
{
"epoch": 1.03,
"grad_norm": 2.756311456596326,
"learning_rate": 1.919154517716744e-05,
"loss": 0.1407,
"step": 7380
},
{
"epoch": 1.04,
"grad_norm": 2.6513725828674386,
"learning_rate": 1.9185447334548534e-05,
"loss": 0.1468,
"step": 7400
},
{
"epoch": 1.04,
"grad_norm": 4.236416595506007,
"learning_rate": 1.9179327557787396e-05,
"loss": 0.1393,
"step": 7420
},
{
"epoch": 1.04,
"grad_norm": 3.0990184275893835,
"learning_rate": 1.917318586149917e-05,
"loss": 0.1485,
"step": 7440
},
{
"epoch": 1.05,
"grad_norm": 2.2486872552215114,
"learning_rate": 1.9167022260351346e-05,
"loss": 0.13,
"step": 7460
},
{
"epoch": 1.05,
"grad_norm": 1.9492044390992067,
"learning_rate": 1.9160836769063744e-05,
"loss": 0.1477,
"step": 7480
},
{
"epoch": 1.05,
"grad_norm": 3.843133023995463,
"learning_rate": 1.9154629402408444e-05,
"loss": 0.1421,
"step": 7500
},
{
"epoch": 1.05,
"grad_norm": 2.559822544563541,
"learning_rate": 1.9148400175209773e-05,
"loss": 0.1427,
"step": 7520
},
{
"epoch": 1.06,
"grad_norm": 6.470849032934646,
"learning_rate": 1.9142149102344265e-05,
"loss": 0.1479,
"step": 7540
},
{
"epoch": 1.06,
"grad_norm": 3.2762712763066717,
"learning_rate": 1.913587619874063e-05,
"loss": 0.1426,
"step": 7560
},
{
"epoch": 1.06,
"grad_norm": 4.007762291428314,
"learning_rate": 1.9129581479379715e-05,
"loss": 0.1447,
"step": 7580
},
{
"epoch": 1.07,
"grad_norm": 2.6372928700764637,
"learning_rate": 1.9123264959294457e-05,
"loss": 0.1494,
"step": 7600
},
{
"epoch": 1.07,
"grad_norm": 2.674816185883692,
"learning_rate": 1.911692665356986e-05,
"loss": 0.1384,
"step": 7620
},
{
"epoch": 1.07,
"grad_norm": 3.2012117952437658,
"learning_rate": 1.9110566577342963e-05,
"loss": 0.1377,
"step": 7640
},
{
"epoch": 1.07,
"grad_norm": 1.8809205993004603,
"learning_rate": 1.9104184745802795e-05,
"loss": 0.1355,
"step": 7660
},
{
"epoch": 1.08,
"grad_norm": 3.0113113464212646,
"learning_rate": 1.9097781174190333e-05,
"loss": 0.1365,
"step": 7680
},
{
"epoch": 1.08,
"grad_norm": 3.2630593474424643,
"learning_rate": 1.909135587779848e-05,
"loss": 0.1337,
"step": 7700
},
{
"epoch": 1.08,
"grad_norm": 11.392702517589797,
"learning_rate": 1.9084908871972026e-05,
"loss": 0.1364,
"step": 7720
},
{
"epoch": 1.09,
"grad_norm": 2.4903594101575486,
"learning_rate": 1.9078440172107598e-05,
"loss": 0.1324,
"step": 7740
},
{
"epoch": 1.09,
"grad_norm": 2.9641453419210446,
"learning_rate": 1.907194979365363e-05,
"loss": 0.1344,
"step": 7760
},
{
"epoch": 1.09,
"grad_norm": 2.768232764848516,
"learning_rate": 1.9065437752110346e-05,
"loss": 0.1434,
"step": 7780
},
{
"epoch": 1.09,
"grad_norm": 3.111667267346916,
"learning_rate": 1.9058904063029682e-05,
"loss": 0.1434,
"step": 7800
},
{
"epoch": 1.1,
"grad_norm": 2.2189052688642668,
"learning_rate": 1.905234874201529e-05,
"loss": 0.1346,
"step": 7820
},
{
"epoch": 1.1,
"grad_norm": 1.938528772222758,
"learning_rate": 1.904577180472248e-05,
"loss": 0.1337,
"step": 7840
},
{
"epoch": 1.1,
"grad_norm": 3.6914431938462164,
"learning_rate": 1.903917326685818e-05,
"loss": 0.1495,
"step": 7860
},
{
"epoch": 1.1,
"grad_norm": 2.693596069469479,
"learning_rate": 1.9032553144180904e-05,
"loss": 0.1389,
"step": 7880
},
{
"epoch": 1.11,
"grad_norm": 2.356678514446313,
"learning_rate": 1.9025911452500718e-05,
"loss": 0.1355,
"step": 7900
},
{
"epoch": 1.11,
"grad_norm": 2.2416126963229313,
"learning_rate": 1.90192482076792e-05,
"loss": 0.1394,
"step": 7920
},
{
"epoch": 1.11,
"grad_norm": 2.407782425838049,
"learning_rate": 1.9012563425629404e-05,
"loss": 0.1444,
"step": 7940
},
{
"epoch": 1.12,
"grad_norm": 2.821450043677631,
"learning_rate": 1.9005857122315802e-05,
"loss": 0.1278,
"step": 7960
},
{
"epoch": 1.12,
"grad_norm": 3.0792341218888803,
"learning_rate": 1.8999129313754287e-05,
"loss": 0.1416,
"step": 7980
},
{
"epoch": 1.12,
"grad_norm": 3.3340412977529343,
"learning_rate": 1.8992380016012084e-05,
"loss": 0.1396,
"step": 8000
},
{
"epoch": 1.12,
"grad_norm": 3.9497092948847667,
"learning_rate": 1.898560924520777e-05,
"loss": 0.1478,
"step": 8020
},
{
"epoch": 1.13,
"grad_norm": 2.5408274927652887,
"learning_rate": 1.8978817017511165e-05,
"loss": 0.15,
"step": 8040
},
{
"epoch": 1.13,
"grad_norm": 2.0665884244115627,
"learning_rate": 1.8972003349143368e-05,
"loss": 0.1412,
"step": 8060
},
{
"epoch": 1.13,
"grad_norm": 3.2381079320231363,
"learning_rate": 1.896516825637666e-05,
"loss": 0.1499,
"step": 8080
},
{
"epoch": 1.14,
"grad_norm": 3.0442534294751304,
"learning_rate": 1.89583117555345e-05,
"loss": 0.1494,
"step": 8100
},
{
"epoch": 1.14,
"grad_norm": 3.2379446188306336,
"learning_rate": 1.895143386299146e-05,
"loss": 0.141,
"step": 8120
},
{
"epoch": 1.14,
"grad_norm": 2.391206059867248,
"learning_rate": 1.8944534595173217e-05,
"loss": 0.1393,
"step": 8140
},
{
"epoch": 1.14,
"grad_norm": 1.5492092207299588,
"learning_rate": 1.8937613968556475e-05,
"loss": 0.1261,
"step": 8160
},
{
"epoch": 1.15,
"grad_norm": 2.7609379594043637,
"learning_rate": 1.893067199966896e-05,
"loss": 0.1318,
"step": 8180
},
{
"epoch": 1.15,
"grad_norm": 3.1261808071845403,
"learning_rate": 1.8923708705089373e-05,
"loss": 0.1444,
"step": 8200
},
{
"epoch": 1.15,
"grad_norm": 3.302597792027258,
"learning_rate": 1.891672410144733e-05,
"loss": 0.1365,
"step": 8220
},
{
"epoch": 1.16,
"grad_norm": 2.642251050152129,
"learning_rate": 1.8909718205423347e-05,
"loss": 0.1354,
"step": 8240
},
{
"epoch": 1.16,
"grad_norm": 2.789807815768997,
"learning_rate": 1.890269103374878e-05,
"loss": 0.1323,
"step": 8260
},
{
"epoch": 1.16,
"grad_norm": 2.5724009461605353,
"learning_rate": 1.889564260320581e-05,
"loss": 0.1545,
"step": 8280
},
{
"epoch": 1.16,
"grad_norm": 3.0712306855541085,
"learning_rate": 1.8888572930627375e-05,
"loss": 0.1434,
"step": 8300
},
{
"epoch": 1.17,
"grad_norm": 3.0627951966306703,
"learning_rate": 1.8881482032897153e-05,
"loss": 0.1398,
"step": 8320
},
{
"epoch": 1.17,
"grad_norm": 2.426100950934894,
"learning_rate": 1.88743699269495e-05,
"loss": 0.1453,
"step": 8340
},
{
"epoch": 1.17,
"grad_norm": 2.724389202013518,
"learning_rate": 1.8867236629769433e-05,
"loss": 0.1308,
"step": 8360
},
{
"epoch": 1.17,
"grad_norm": 1.7926183952484913,
"learning_rate": 1.8860082158392566e-05,
"loss": 0.1398,
"step": 8380
},
{
"epoch": 1.18,
"grad_norm": 2.568057291229617,
"learning_rate": 1.885290652990509e-05,
"loss": 0.1534,
"step": 8400
},
{
"epoch": 1.18,
"grad_norm": 2.495273891238668,
"learning_rate": 1.8845709761443723e-05,
"loss": 0.1388,
"step": 8420
},
{
"epoch": 1.18,
"grad_norm": 2.7499667078950703,
"learning_rate": 1.883849187019566e-05,
"loss": 0.1281,
"step": 8440
},
{
"epoch": 1.19,
"grad_norm": 3.2641865605817806,
"learning_rate": 1.8831252873398554e-05,
"loss": 0.1336,
"step": 8460
},
{
"epoch": 1.19,
"grad_norm": 2.1042885949591716,
"learning_rate": 1.882399278834045e-05,
"loss": 0.1431,
"step": 8480
},
{
"epoch": 1.19,
"grad_norm": 2.6052393318985008,
"learning_rate": 1.8816711632359764e-05,
"loss": 0.1386,
"step": 8500
},
{
"epoch": 1.19,
"grad_norm": 4.3872587393047,
"learning_rate": 1.8809409422845225e-05,
"loss": 0.143,
"step": 8520
},
{
"epoch": 1.2,
"grad_norm": 2.7857596023422833,
"learning_rate": 1.8802086177235853e-05,
"loss": 0.1512,
"step": 8540
},
{
"epoch": 1.2,
"grad_norm": 2.698442405426744,
"learning_rate": 1.8794741913020897e-05,
"loss": 0.1495,
"step": 8560
},
{
"epoch": 1.2,
"grad_norm": 6.4120894451220884,
"learning_rate": 1.87873766477398e-05,
"loss": 0.1369,
"step": 8580
},
{
"epoch": 1.21,
"grad_norm": 6.544595471754997,
"learning_rate": 1.8779990398982174e-05,
"loss": 0.1304,
"step": 8600
},
{
"epoch": 1.21,
"grad_norm": 2.4341265491634836,
"learning_rate": 1.877258318438772e-05,
"loss": 0.1377,
"step": 8620
},
{
"epoch": 1.21,
"grad_norm": 2.711913710839012,
"learning_rate": 1.8765155021646237e-05,
"loss": 0.1574,
"step": 8640
},
{
"epoch": 1.21,
"grad_norm": 2.5531671415514015,
"learning_rate": 1.8757705928497532e-05,
"loss": 0.1362,
"step": 8660
},
{
"epoch": 1.22,
"grad_norm": 2.762561804308771,
"learning_rate": 1.8750235922731405e-05,
"loss": 0.1251,
"step": 8680
},
{
"epoch": 1.22,
"grad_norm": 2.584772314230473,
"learning_rate": 1.8742745022187597e-05,
"loss": 0.139,
"step": 8700
},
{
"epoch": 1.22,
"grad_norm": 2.701585011253313,
"learning_rate": 1.8735233244755747e-05,
"loss": 0.1355,
"step": 8720
},
{
"epoch": 1.23,
"grad_norm": 2.648877492719425,
"learning_rate": 1.872770060837536e-05,
"loss": 0.1486,
"step": 8740
},
{
"epoch": 1.23,
"grad_norm": 2.438552531386958,
"learning_rate": 1.8720147131035752e-05,
"loss": 0.1395,
"step": 8760
},
{
"epoch": 1.23,
"grad_norm": 2.0825974372700733,
"learning_rate": 1.8712572830776004e-05,
"loss": 0.1299,
"step": 8780
},
{
"epoch": 1.23,
"grad_norm": 2.422483503130003,
"learning_rate": 1.8704977725684942e-05,
"loss": 0.1231,
"step": 8800
},
{
"epoch": 1.24,
"grad_norm": 2.197873179892911,
"learning_rate": 1.8697361833901055e-05,
"loss": 0.1298,
"step": 8820
},
{
"epoch": 1.24,
"grad_norm": 11.94520200549779,
"learning_rate": 1.8689725173612494e-05,
"loss": 0.1404,
"step": 8840
},
{
"epoch": 1.24,
"grad_norm": 5.2570143571395995,
"learning_rate": 1.8682067763057005e-05,
"loss": 0.1372,
"step": 8860
},
{
"epoch": 1.24,
"grad_norm": 2.995419979803786,
"learning_rate": 1.8674389620521885e-05,
"loss": 0.1468,
"step": 8880
},
{
"epoch": 1.25,
"grad_norm": 2.071293551643827,
"learning_rate": 1.8666690764343942e-05,
"loss": 0.149,
"step": 8900
},
{
"epoch": 1.25,
"grad_norm": 2.73550104116184,
"learning_rate": 1.8658971212909456e-05,
"loss": 0.1368,
"step": 8920
},
{
"epoch": 1.25,
"grad_norm": 3.2514442755980513,
"learning_rate": 1.8651230984654128e-05,
"loss": 0.1393,
"step": 8940
},
{
"epoch": 1.26,
"grad_norm": 2.8867692955867432,
"learning_rate": 1.8643470098063047e-05,
"loss": 0.1367,
"step": 8960
},
{
"epoch": 1.26,
"grad_norm": 1.746842669629685,
"learning_rate": 1.8635688571670626e-05,
"loss": 0.1421,
"step": 8980
},
{
"epoch": 1.26,
"grad_norm": 2.4432616972819408,
"learning_rate": 1.8627886424060574e-05,
"loss": 0.1438,
"step": 9000
},
{
"epoch": 1.26,
"grad_norm": 2.35822242727653,
"learning_rate": 1.862006367386585e-05,
"loss": 0.1504,
"step": 9020
},
{
"epoch": 1.27,
"grad_norm": 2.599251301879498,
"learning_rate": 1.8612220339768618e-05,
"loss": 0.1383,
"step": 9040
},
{
"epoch": 1.27,
"grad_norm": 2.1269373196942025,
"learning_rate": 1.8604356440500183e-05,
"loss": 0.1397,
"step": 9060
},
{
"epoch": 1.27,
"grad_norm": 2.993386608519107,
"learning_rate": 1.8596471994840997e-05,
"loss": 0.1343,
"step": 9080
},
{
"epoch": 1.28,
"grad_norm": 2.122458789726414,
"learning_rate": 1.8588567021620543e-05,
"loss": 0.129,
"step": 9100
},
{
"epoch": 1.28,
"grad_norm": 3.987398683223458,
"learning_rate": 1.858064153971735e-05,
"loss": 0.1363,
"step": 9120
},
{
"epoch": 1.28,
"grad_norm": 2.3148730705842726,
"learning_rate": 1.8572695568058928e-05,
"loss": 0.142,
"step": 9140
},
{
"epoch": 1.28,
"grad_norm": 4.748410410676678,
"learning_rate": 1.8564729125621703e-05,
"loss": 0.15,
"step": 9160
},
{
"epoch": 1.29,
"grad_norm": 2.85836023553435,
"learning_rate": 1.8556742231430997e-05,
"loss": 0.1387,
"step": 9180
},
{
"epoch": 1.29,
"grad_norm": 2.099542717873591,
"learning_rate": 1.8548734904560983e-05,
"loss": 0.1515,
"step": 9200
},
{
"epoch": 1.29,
"grad_norm": 3.1016273780013894,
"learning_rate": 1.8540707164134622e-05,
"loss": 0.1428,
"step": 9220
},
{
"epoch": 1.3,
"grad_norm": 2.227753651880202,
"learning_rate": 1.8532659029323628e-05,
"loss": 0.1449,
"step": 9240
},
{
"epoch": 1.3,
"grad_norm": 2.755045337338574,
"learning_rate": 1.8524590519348417e-05,
"loss": 0.1371,
"step": 9260
},
{
"epoch": 1.3,
"grad_norm": 3.807800147686546,
"learning_rate": 1.8516501653478076e-05,
"loss": 0.1431,
"step": 9280
},
{
"epoch": 1.3,
"grad_norm": 1.9715523529879975,
"learning_rate": 1.850839245103029e-05,
"loss": 0.141,
"step": 9300
},
{
"epoch": 1.31,
"grad_norm": 2.5911895799378453,
"learning_rate": 1.850026293137133e-05,
"loss": 0.1373,
"step": 9320
},
{
"epoch": 1.31,
"grad_norm": 1.7898255432614876,
"learning_rate": 1.8492113113915964e-05,
"loss": 0.1365,
"step": 9340
},
{
"epoch": 1.31,
"grad_norm": 2.27885488949853,
"learning_rate": 1.848394301812746e-05,
"loss": 0.1334,
"step": 9360
},
{
"epoch": 1.32,
"grad_norm": 2.3597330996351413,
"learning_rate": 1.8475752663517498e-05,
"loss": 0.1468,
"step": 9380
},
{
"epoch": 1.32,
"grad_norm": 3.16996158708951,
"learning_rate": 1.8467542069646143e-05,
"loss": 0.1355,
"step": 9400
},
{
"epoch": 1.32,
"grad_norm": 3.123146117579664,
"learning_rate": 1.84593112561218e-05,
"loss": 0.1413,
"step": 9420
},
{
"epoch": 1.32,
"grad_norm": 4.168978596472018,
"learning_rate": 1.845106024260116e-05,
"loss": 0.1282,
"step": 9440
},
{
"epoch": 1.33,
"grad_norm": 2.42991426151806,
"learning_rate": 1.844278904878915e-05,
"loss": 0.1355,
"step": 9460
},
{
"epoch": 1.33,
"grad_norm": 2.707913693905128,
"learning_rate": 1.84344976944389e-05,
"loss": 0.1582,
"step": 9480
},
{
"epoch": 1.33,
"grad_norm": 3.1479488779301237,
"learning_rate": 1.8426186199351686e-05,
"loss": 0.1333,
"step": 9500
},
{
"epoch": 1.33,
"grad_norm": 4.786796195386024,
"learning_rate": 1.841785458337687e-05,
"loss": 0.1278,
"step": 9520
},
{
"epoch": 1.34,
"grad_norm": 2.090372578388717,
"learning_rate": 1.840950286641188e-05,
"loss": 0.1437,
"step": 9540
},
{
"epoch": 1.34,
"grad_norm": 2.046449223853787,
"learning_rate": 1.8401131068402153e-05,
"loss": 0.1381,
"step": 9560
},
{
"epoch": 1.34,
"grad_norm": 3.699356672702717,
"learning_rate": 1.839273920934107e-05,
"loss": 0.1256,
"step": 9580
},
{
"epoch": 1.35,
"grad_norm": 2.1840007586914534,
"learning_rate": 1.8384327309269923e-05,
"loss": 0.1448,
"step": 9600
},
{
"epoch": 1.35,
"grad_norm": 2.396410156692531,
"learning_rate": 1.8375895388277878e-05,
"loss": 0.1372,
"step": 9620
},
{
"epoch": 1.35,
"grad_norm": 3.9533054476478724,
"learning_rate": 1.8367443466501906e-05,
"loss": 0.1397,
"step": 9640
},
{
"epoch": 1.35,
"grad_norm": 2.6698289142043965,
"learning_rate": 1.8358971564126734e-05,
"loss": 0.1473,
"step": 9660
},
{
"epoch": 1.36,
"grad_norm": 3.143115977367074,
"learning_rate": 1.8350479701384828e-05,
"loss": 0.1415,
"step": 9680
},
{
"epoch": 1.36,
"grad_norm": 3.04930088598182,
"learning_rate": 1.834196789855631e-05,
"loss": 0.127,
"step": 9700
},
{
"epoch": 1.36,
"grad_norm": 2.013659561371177,
"learning_rate": 1.8333436175968914e-05,
"loss": 0.1523,
"step": 9720
},
{
"epoch": 1.37,
"grad_norm": 2.083768481268159,
"learning_rate": 1.8324884553997966e-05,
"loss": 0.1383,
"step": 9740
},
{
"epoch": 1.37,
"grad_norm": 2.6418929808351965,
"learning_rate": 1.8316313053066304e-05,
"loss": 0.1408,
"step": 9760
},
{
"epoch": 1.37,
"grad_norm": 3.665927696901052,
"learning_rate": 1.8307721693644237e-05,
"loss": 0.1468,
"step": 9780
},
{
"epoch": 1.37,
"grad_norm": 1.9664353584860104,
"learning_rate": 1.8299110496249512e-05,
"loss": 0.1311,
"step": 9800
},
{
"epoch": 1.38,
"grad_norm": 2.8218143933991615,
"learning_rate": 1.8290479481447242e-05,
"loss": 0.1381,
"step": 9820
},
{
"epoch": 1.38,
"grad_norm": 2.2011951624697046,
"learning_rate": 1.828182866984987e-05,
"loss": 0.1319,
"step": 9840
},
{
"epoch": 1.38,
"grad_norm": 3.0649843065566382,
"learning_rate": 1.827315808211711e-05,
"loss": 0.125,
"step": 9860
},
{
"epoch": 1.39,
"grad_norm": 2.712357294925747,
"learning_rate": 1.826446773895593e-05,
"loss": 0.1421,
"step": 9880
},
{
"epoch": 1.39,
"grad_norm": 2.7613981889132098,
"learning_rate": 1.8255757661120455e-05,
"loss": 0.1494,
"step": 9900
},
{
"epoch": 1.39,
"grad_norm": 1.8936891552306527,
"learning_rate": 1.8247027869411944e-05,
"loss": 0.1433,
"step": 9920
},
{
"epoch": 1.39,
"grad_norm": 3.9453841703344783,
"learning_rate": 1.8238278384678733e-05,
"loss": 0.1356,
"step": 9940
},
{
"epoch": 1.4,
"grad_norm": 7.036388638760909,
"learning_rate": 1.8229509227816206e-05,
"loss": 0.1381,
"step": 9960
},
{
"epoch": 1.4,
"grad_norm": 2.6932148165010985,
"learning_rate": 1.8220720419766707e-05,
"loss": 0.125,
"step": 9980
},
{
"epoch": 1.4,
"grad_norm": 2.6453561327468327,
"learning_rate": 1.821191198151952e-05,
"loss": 0.1502,
"step": 10000
},
{
"epoch": 1.4,
"grad_norm": 2.1235897770302006,
"learning_rate": 1.820308393411081e-05,
"loss": 0.1279,
"step": 10020
},
{
"epoch": 1.41,
"grad_norm": 1.435776880224214,
"learning_rate": 1.8194236298623567e-05,
"loss": 0.1172,
"step": 10040
},
{
"epoch": 1.41,
"grad_norm": 1.7288853826353796,
"learning_rate": 1.8185369096187576e-05,
"loss": 0.1231,
"step": 10060
},
{
"epoch": 1.41,
"grad_norm": 1.5077340693656098,
"learning_rate": 1.8176482347979324e-05,
"loss": 0.1204,
"step": 10080
},
{
"epoch": 1.42,
"grad_norm": 1.5100641387302804,
"learning_rate": 1.8167576075222007e-05,
"loss": 0.1154,
"step": 10100
},
{
"epoch": 1.42,
"grad_norm": 1.7400330363897294,
"learning_rate": 1.8158650299185433e-05,
"loss": 0.1282,
"step": 10120
},
{
"epoch": 1.42,
"grad_norm": 1.9280372389445906,
"learning_rate": 1.814970504118598e-05,
"loss": 0.1177,
"step": 10140
},
{
"epoch": 1.42,
"grad_norm": 1.6902178318103636,
"learning_rate": 1.8140740322586576e-05,
"loss": 0.122,
"step": 10160
},
{
"epoch": 1.43,
"grad_norm": 1.567397263124585,
"learning_rate": 1.8131756164796604e-05,
"loss": 0.1157,
"step": 10180
},
{
"epoch": 1.43,
"grad_norm": 1.4493628581905045,
"learning_rate": 1.8122752589271877e-05,
"loss": 0.1183,
"step": 10200
},
{
"epoch": 1.43,
"grad_norm": 1.7435456870645718,
"learning_rate": 1.8113729617514583e-05,
"loss": 0.1215,
"step": 10220
},
{
"epoch": 1.44,
"grad_norm": 1.556704568925633,
"learning_rate": 1.8104687271073234e-05,
"loss": 0.115,
"step": 10240
},
{
"epoch": 1.44,
"grad_norm": 10.929147899251383,
"learning_rate": 1.8095625571542605e-05,
"loss": 0.1138,
"step": 10260
},
{
"epoch": 1.44,
"grad_norm": 1.9287889373072822,
"learning_rate": 1.8086544540563695e-05,
"loss": 0.122,
"step": 10280
},
{
"epoch": 1.44,
"grad_norm": 3.982492757767907,
"learning_rate": 1.8077444199823672e-05,
"loss": 0.1158,
"step": 10300
},
{
"epoch": 1.45,
"grad_norm": 1.6429950126112538,
"learning_rate": 1.8068324571055817e-05,
"loss": 0.1119,
"step": 10320
},
{
"epoch": 1.45,
"grad_norm": 1.7704667403795766,
"learning_rate": 1.8059185676039466e-05,
"loss": 0.1139,
"step": 10340
},
{
"epoch": 1.45,
"grad_norm": 1.468087960235158,
"learning_rate": 1.8050027536599982e-05,
"loss": 0.1143,
"step": 10360
},
{
"epoch": 1.46,
"grad_norm": 1.2763392613923261,
"learning_rate": 1.8040850174608675e-05,
"loss": 0.1153,
"step": 10380
},
{
"epoch": 1.46,
"grad_norm": 1.3699321118818848,
"learning_rate": 1.8031653611982765e-05,
"loss": 0.116,
"step": 10400
},
{
"epoch": 1.46,
"grad_norm": 1.5674554445544704,
"learning_rate": 1.8022437870685333e-05,
"loss": 0.1091,
"step": 10420
},
{
"epoch": 1.46,
"grad_norm": 1.5276199442320406,
"learning_rate": 1.801320297272525e-05,
"loss": 0.1204,
"step": 10440
},
{
"epoch": 1.47,
"grad_norm": 1.6746576984285857,
"learning_rate": 1.8003948940157156e-05,
"loss": 0.1091,
"step": 10460
},
{
"epoch": 1.47,
"grad_norm": 2.986207090111833,
"learning_rate": 1.7994675795081358e-05,
"loss": 0.1087,
"step": 10480
},
{
"epoch": 1.47,
"grad_norm": 1.36546142611082,
"learning_rate": 1.7985383559643836e-05,
"loss": 0.1121,
"step": 10500
},
{
"epoch": 1.47,
"grad_norm": 1.5098551616298637,
"learning_rate": 1.797607225603615e-05,
"loss": 0.1138,
"step": 10520
},
{
"epoch": 1.48,
"grad_norm": 1.8344336011322107,
"learning_rate": 1.7966741906495396e-05,
"loss": 0.1154,
"step": 10540
},
{
"epoch": 1.48,
"grad_norm": 8.018871936183185,
"learning_rate": 1.7957392533304154e-05,
"loss": 0.1142,
"step": 10560
},
{
"epoch": 1.48,
"grad_norm": 1.7230257429523406,
"learning_rate": 1.7948024158790445e-05,
"loss": 0.1234,
"step": 10580
},
{
"epoch": 1.49,
"grad_norm": 1.2925443384914297,
"learning_rate": 1.793863680532766e-05,
"loss": 0.1208,
"step": 10600
},
{
"epoch": 1.49,
"grad_norm": 1.415894896523647,
"learning_rate": 1.792923049533452e-05,
"loss": 0.1145,
"step": 10620
},
{
"epoch": 1.49,
"grad_norm": 1.6520892008051478,
"learning_rate": 1.7919805251275015e-05,
"loss": 0.1094,
"step": 10640
},
{
"epoch": 1.49,
"grad_norm": 1.2677552462412662,
"learning_rate": 1.7910361095658352e-05,
"loss": 0.1141,
"step": 10660
},
{
"epoch": 1.5,
"grad_norm": 1.5377786771469084,
"learning_rate": 1.790089805103891e-05,
"loss": 0.11,
"step": 10680
},
{
"epoch": 1.5,
"grad_norm": 2.238864256873464,
"learning_rate": 1.7891416140016162e-05,
"loss": 0.1145,
"step": 10700
},
{
"epoch": 1.5,
"grad_norm": 1.9123861017157395,
"learning_rate": 1.788191538523466e-05,
"loss": 0.1143,
"step": 10720
},
{
"epoch": 1.51,
"grad_norm": 1.744441878903413,
"learning_rate": 1.787239580938394e-05,
"loss": 0.1169,
"step": 10740
},
{
"epoch": 1.51,
"grad_norm": 1.0123552360183345,
"learning_rate": 1.786285743519849e-05,
"loss": 0.1193,
"step": 10760
},
{
"epoch": 1.51,
"grad_norm": 1.6439464553225356,
"learning_rate": 1.7853300285457703e-05,
"loss": 0.1115,
"step": 10780
},
{
"epoch": 1.51,
"grad_norm": 1.0766077284039297,
"learning_rate": 1.7843724382985794e-05,
"loss": 0.1131,
"step": 10800
},
{
"epoch": 1.52,
"grad_norm": 1.2612340604381689,
"learning_rate": 1.7834129750651777e-05,
"loss": 0.1126,
"step": 10820
},
{
"epoch": 1.52,
"grad_norm": 2.0146242954806532,
"learning_rate": 1.7824516411369387e-05,
"loss": 0.1058,
"step": 10840
},
{
"epoch": 1.52,
"grad_norm": 1.2684292279725826,
"learning_rate": 1.7814884388097043e-05,
"loss": 0.1184,
"step": 10860
},
{
"epoch": 1.53,
"grad_norm": 1.571691183031815,
"learning_rate": 1.780523370383778e-05,
"loss": 0.1098,
"step": 10880
},
{
"epoch": 1.53,
"grad_norm": 2.111364937742784,
"learning_rate": 1.7795564381639193e-05,
"loss": 0.1148,
"step": 10900
},
{
"epoch": 1.53,
"grad_norm": 1.2381626397255368,
"learning_rate": 1.7785876444593407e-05,
"loss": 0.1156,
"step": 10920
},
{
"epoch": 1.53,
"grad_norm": 1.5119570364423744,
"learning_rate": 1.777616991583698e-05,
"loss": 0.1125,
"step": 10940
},
{
"epoch": 1.54,
"grad_norm": 1.3882498871239757,
"learning_rate": 1.776644481855088e-05,
"loss": 0.1111,
"step": 10960
},
{
"epoch": 1.54,
"grad_norm": 2.967835938127249,
"learning_rate": 1.7756701175960424e-05,
"loss": 0.112,
"step": 10980
},
{
"epoch": 1.54,
"grad_norm": 1.46013223415033,
"learning_rate": 1.7746939011335213e-05,
"loss": 0.1071,
"step": 11000
},
{
"epoch": 1.54,
"grad_norm": 1.6895164169186756,
"learning_rate": 1.7737158347989085e-05,
"loss": 0.1145,
"step": 11020
},
{
"epoch": 1.55,
"grad_norm": 1.4115504508088004,
"learning_rate": 1.7727359209280056e-05,
"loss": 0.1061,
"step": 11040
},
{
"epoch": 1.55,
"grad_norm": 1.4420350726778908,
"learning_rate": 1.7717541618610265e-05,
"loss": 0.1111,
"step": 11060
},
{
"epoch": 1.55,
"grad_norm": 1.3960962142879894,
"learning_rate": 1.770770559942591e-05,
"loss": 0.1063,
"step": 11080
},
{
"epoch": 1.56,
"grad_norm": 1.5112664386369516,
"learning_rate": 1.769785117521721e-05,
"loss": 0.1092,
"step": 11100
},
{
"epoch": 1.56,
"grad_norm": 1.5397154686386814,
"learning_rate": 1.768797836951834e-05,
"loss": 0.1226,
"step": 11120
},
{
"epoch": 1.56,
"grad_norm": 1.3786648114777287,
"learning_rate": 1.767808720590736e-05,
"loss": 0.1136,
"step": 11140
},
{
"epoch": 1.56,
"grad_norm": 1.2388738423637375,
"learning_rate": 1.7668177708006186e-05,
"loss": 0.1147,
"step": 11160
},
{
"epoch": 1.57,
"grad_norm": 1.4430222795158352,
"learning_rate": 1.765824989948051e-05,
"loss": 0.112,
"step": 11180
},
{
"epoch": 1.57,
"grad_norm": 1.2041265109185475,
"learning_rate": 1.7648303804039754e-05,
"loss": 0.1098,
"step": 11200
},
{
"epoch": 1.57,
"grad_norm": 1.8045111896213413,
"learning_rate": 1.763833944543702e-05,
"loss": 0.1094,
"step": 11220
},
{
"epoch": 1.58,
"grad_norm": 1.3852779586688524,
"learning_rate": 1.7628356847469022e-05,
"loss": 0.1168,
"step": 11240
},
{
"epoch": 1.58,
"grad_norm": 1.2992496580850497,
"learning_rate": 1.7618356033976033e-05,
"loss": 0.1012,
"step": 11260
},
{
"epoch": 1.58,
"grad_norm": 1.2356511533031083,
"learning_rate": 1.7608337028841828e-05,
"loss": 0.1106,
"step": 11280
},
{
"epoch": 1.58,
"grad_norm": 2.6902483704215236,
"learning_rate": 1.759829985599362e-05,
"loss": 0.1142,
"step": 11300
},
{
"epoch": 1.59,
"grad_norm": 1.4195172390537656,
"learning_rate": 1.758824453940201e-05,
"loss": 0.1039,
"step": 11320
},
{
"epoch": 1.59,
"grad_norm": 2.604728170259304,
"learning_rate": 1.7578171103080953e-05,
"loss": 0.1152,
"step": 11340
},
{
"epoch": 1.59,
"grad_norm": 1.2178920758418774,
"learning_rate": 1.756807957108765e-05,
"loss": 0.1077,
"step": 11360
},
{
"epoch": 1.6,
"grad_norm": 1.4334971216051138,
"learning_rate": 1.755796996752253e-05,
"loss": 0.1049,
"step": 11380
},
{
"epoch": 1.6,
"grad_norm": 1.3758639309163114,
"learning_rate": 1.754784231652917e-05,
"loss": 0.1156,
"step": 11400
},
{
"epoch": 1.6,
"grad_norm": 1.1826910221100024,
"learning_rate": 1.7537696642294268e-05,
"loss": 0.1067,
"step": 11420
},
{
"epoch": 1.6,
"grad_norm": 1.7739652948995182,
"learning_rate": 1.7527532969047544e-05,
"loss": 0.122,
"step": 11440
},
{
"epoch": 1.61,
"grad_norm": 1.9689234702570657,
"learning_rate": 1.751735132106171e-05,
"loss": 0.1199,
"step": 11460
},
{
"epoch": 1.61,
"grad_norm": 1.6716582255785815,
"learning_rate": 1.7507151722652417e-05,
"loss": 0.1169,
"step": 11480
},
{
"epoch": 1.61,
"grad_norm": 4.663822851621681,
"learning_rate": 1.7496934198178155e-05,
"loss": 0.1136,
"step": 11500
},
{
"epoch": 1.62,
"grad_norm": 1.831089648103746,
"learning_rate": 1.7486698772040266e-05,
"loss": 0.1204,
"step": 11520
},
{
"epoch": 1.62,
"grad_norm": 1.3153019016787797,
"learning_rate": 1.7476445468682802e-05,
"loss": 0.1119,
"step": 11540
},
{
"epoch": 1.62,
"grad_norm": 1.1536774756553378,
"learning_rate": 1.746617431259254e-05,
"loss": 0.1093,
"step": 11560
},
{
"epoch": 1.62,
"grad_norm": 1.4306398982067656,
"learning_rate": 1.7455885328298877e-05,
"loss": 0.1101,
"step": 11580
},
{
"epoch": 1.63,
"grad_norm": 1.4822167215637623,
"learning_rate": 1.7445578540373797e-05,
"loss": 0.117,
"step": 11600
},
{
"epoch": 1.63,
"grad_norm": 1.2700435602422147,
"learning_rate": 1.7435253973431795e-05,
"loss": 0.1094,
"step": 11620
},
{
"epoch": 1.63,
"grad_norm": 1.4263562022996301,
"learning_rate": 1.7424911652129826e-05,
"loss": 0.1145,
"step": 11640
},
{
"epoch": 1.63,
"grad_norm": 1.6292901993118378,
"learning_rate": 1.741455160116725e-05,
"loss": 0.1108,
"step": 11660
},
{
"epoch": 1.64,
"grad_norm": 1.3657575704444023,
"learning_rate": 1.740417384528576e-05,
"loss": 0.1105,
"step": 11680
},
{
"epoch": 1.64,
"grad_norm": 1.412593267956803,
"learning_rate": 1.7393778409269345e-05,
"loss": 0.119,
"step": 11700
},
{
"epoch": 1.64,
"grad_norm": 1.5907997799760736,
"learning_rate": 1.73833653179442e-05,
"loss": 0.114,
"step": 11720
},
{
"epoch": 1.65,
"grad_norm": 1.2927153186266116,
"learning_rate": 1.737293459617871e-05,
"loss": 0.1088,
"step": 11740
},
{
"epoch": 1.65,
"grad_norm": 1.311938165577001,
"learning_rate": 1.7362486268883335e-05,
"loss": 0.1046,
"step": 11760
},
{
"epoch": 1.65,
"grad_norm": 1.6905621190813724,
"learning_rate": 1.73520203610106e-05,
"loss": 0.1116,
"step": 11780
},
{
"epoch": 1.65,
"grad_norm": 1.4808323396082979,
"learning_rate": 1.734153689755501e-05,
"loss": 0.1075,
"step": 11800
},
{
"epoch": 1.66,
"grad_norm": 1.4392263578057085,
"learning_rate": 1.7331035903552992e-05,
"loss": 0.1089,
"step": 11820
},
{
"epoch": 1.66,
"grad_norm": 1.8030743676122682,
"learning_rate": 1.7320517404082845e-05,
"loss": 0.1075,
"step": 11840
},
{
"epoch": 1.66,
"grad_norm": 1.3012140144114108,
"learning_rate": 1.730998142426467e-05,
"loss": 0.1052,
"step": 11860
},
{
"epoch": 1.67,
"grad_norm": 1.5158754122225413,
"learning_rate": 1.7299427989260326e-05,
"loss": 0.1124,
"step": 11880
},
{
"epoch": 1.67,
"grad_norm": 1.5784614884104387,
"learning_rate": 1.7288857124273332e-05,
"loss": 0.1044,
"step": 11900
},
{
"epoch": 1.67,
"grad_norm": 1.091339998412087,
"learning_rate": 1.7278268854548863e-05,
"loss": 0.1086,
"step": 11920
},
{
"epoch": 1.67,
"grad_norm": 1.5574412279383307,
"learning_rate": 1.7267663205373637e-05,
"loss": 0.111,
"step": 11940
},
{
"epoch": 1.68,
"grad_norm": 1.965375217887305,
"learning_rate": 1.7257040202075895e-05,
"loss": 0.1036,
"step": 11960
},
{
"epoch": 1.68,
"grad_norm": 2.56450379344104,
"learning_rate": 1.724639987002531e-05,
"loss": 0.1099,
"step": 11980
},
{
"epoch": 1.68,
"grad_norm": 1.8766284229092605,
"learning_rate": 1.723574223463294e-05,
"loss": 0.1303,
"step": 12000
},
{
"epoch": 1.69,
"grad_norm": 1.5159752824424013,
"learning_rate": 1.722506732135118e-05,
"loss": 0.1032,
"step": 12020
},
{
"epoch": 1.69,
"grad_norm": 1.590345448808343,
"learning_rate": 1.7214375155673665e-05,
"loss": 0.1119,
"step": 12040
},
{
"epoch": 1.69,
"grad_norm": 1.4596903980890314,
"learning_rate": 1.7203665763135256e-05,
"loss": 0.1084,
"step": 12060
},
{
"epoch": 1.69,
"grad_norm": 1.495012973614284,
"learning_rate": 1.7192939169311938e-05,
"loss": 0.1038,
"step": 12080
},
{
"epoch": 1.7,
"grad_norm": 1.5820568694877744,
"learning_rate": 1.7182195399820784e-05,
"loss": 0.1158,
"step": 12100
},
{
"epoch": 1.7,
"grad_norm": 2.0147080812487266,
"learning_rate": 1.7171434480319878e-05,
"loss": 0.1147,
"step": 12120
},
{
"epoch": 1.7,
"grad_norm": 1.3133909289725714,
"learning_rate": 1.7160656436508275e-05,
"loss": 0.1134,
"step": 12140
},
{
"epoch": 1.7,
"grad_norm": 1.6866021063414325,
"learning_rate": 1.714986129412591e-05,
"loss": 0.1027,
"step": 12160
},
{
"epoch": 1.71,
"grad_norm": 2.414428485669963,
"learning_rate": 1.713904907895356e-05,
"loss": 0.1162,
"step": 12180
},
{
"epoch": 1.71,
"grad_norm": 1.2898660640530863,
"learning_rate": 1.7128219816812782e-05,
"loss": 0.1101,
"step": 12200
},
{
"epoch": 1.71,
"grad_norm": 1.1065570340564246,
"learning_rate": 1.7117373533565833e-05,
"loss": 0.1093,
"step": 12220
},
{
"epoch": 1.72,
"grad_norm": 1.4323076097005671,
"learning_rate": 1.710651025511562e-05,
"loss": 0.1138,
"step": 12240
},
{
"epoch": 1.72,
"grad_norm": 1.8418493657808765,
"learning_rate": 1.7095630007405648e-05,
"loss": 0.1144,
"step": 12260
},
{
"epoch": 1.72,
"grad_norm": 1.4285044126459157,
"learning_rate": 1.7084732816419937e-05,
"loss": 0.1003,
"step": 12280
},
{
"epoch": 1.72,
"grad_norm": 1.6012121794356293,
"learning_rate": 1.7073818708182978e-05,
"loss": 0.1089,
"step": 12300
},
{
"epoch": 1.73,
"grad_norm": 1.6497117310912452,
"learning_rate": 1.7062887708759653e-05,
"loss": 0.1071,
"step": 12320
},
{
"epoch": 1.73,
"grad_norm": 1.2516425308375894,
"learning_rate": 1.70519398442552e-05,
"loss": 0.1158,
"step": 12340
},
{
"epoch": 1.73,
"grad_norm": 1.5672025487753456,
"learning_rate": 1.7040975140815118e-05,
"loss": 0.1043,
"step": 12360
},
{
"epoch": 1.74,
"grad_norm": 1.1161205825880594,
"learning_rate": 1.7029993624625133e-05,
"loss": 0.1034,
"step": 12380
},
{
"epoch": 1.74,
"grad_norm": 1.7523116103280694,
"learning_rate": 1.7018995321911108e-05,
"loss": 0.1101,
"step": 12400
},
{
"epoch": 1.74,
"grad_norm": 1.4726475075378764,
"learning_rate": 1.7007980258939008e-05,
"loss": 0.1086,
"step": 12420
},
{
"epoch": 1.74,
"grad_norm": 1.7489560283064822,
"learning_rate": 1.699694846201482e-05,
"loss": 0.1121,
"step": 12440
},
{
"epoch": 1.75,
"grad_norm": 0.9707718331718185,
"learning_rate": 1.6985899957484502e-05,
"loss": 0.1186,
"step": 12460
},
{
"epoch": 1.75,
"grad_norm": 1.4412762342912322,
"learning_rate": 1.6974834771733895e-05,
"loss": 0.1163,
"step": 12480
},
{
"epoch": 1.75,
"grad_norm": 1.688102897038925,
"learning_rate": 1.69637529311887e-05,
"loss": 0.1057,
"step": 12500
},
{
"epoch": 1.76,
"grad_norm": 1.4797675653888702,
"learning_rate": 1.6952654462314376e-05,
"loss": 0.1179,
"step": 12520
},
{
"epoch": 1.76,
"grad_norm": 1.200870283768574,
"learning_rate": 1.6941539391616095e-05,
"loss": 0.1164,
"step": 12540
},
{
"epoch": 1.76,
"grad_norm": 1.3604519403781055,
"learning_rate": 1.693040774563869e-05,
"loss": 0.1012,
"step": 12560
},
{
"epoch": 1.76,
"grad_norm": 1.2805964501076283,
"learning_rate": 1.6919259550966564e-05,
"loss": 0.1138,
"step": 12580
},
{
"epoch": 1.77,
"grad_norm": 2.0391798899099327,
"learning_rate": 1.6908094834223653e-05,
"loss": 0.116,
"step": 12600
},
{
"epoch": 1.77,
"grad_norm": 1.8528360743059575,
"learning_rate": 1.6896913622073342e-05,
"loss": 0.1182,
"step": 12620
},
{
"epoch": 1.77,
"grad_norm": 1.7966655443267054,
"learning_rate": 1.688571594121841e-05,
"loss": 0.1072,
"step": 12640
},
{
"epoch": 1.77,
"grad_norm": 1.5052086677242906,
"learning_rate": 1.6874501818400976e-05,
"loss": 0.1104,
"step": 12660
},
{
"epoch": 1.78,
"grad_norm": 1.417478534461951,
"learning_rate": 1.6863271280402413e-05,
"loss": 0.1107,
"step": 12680
},
{
"epoch": 1.78,
"grad_norm": 1.6827628437391418,
"learning_rate": 1.6852024354043305e-05,
"loss": 0.1138,
"step": 12700
},
{
"epoch": 1.78,
"grad_norm": 2.7578858103747526,
"learning_rate": 1.6840761066183373e-05,
"loss": 0.1087,
"step": 12720
},
{
"epoch": 1.79,
"grad_norm": 1.870018381242629,
"learning_rate": 1.682948144372141e-05,
"loss": 0.0992,
"step": 12740
},
{
"epoch": 1.79,
"grad_norm": 1.764139951858943,
"learning_rate": 1.6818185513595222e-05,
"loss": 0.1083,
"step": 12760
},
{
"epoch": 1.79,
"grad_norm": 1.4899414094503753,
"learning_rate": 1.680687330278156e-05,
"loss": 0.1111,
"step": 12780
},
{
"epoch": 1.79,
"grad_norm": 1.412289092260433,
"learning_rate": 1.679554483829605e-05,
"loss": 0.0992,
"step": 12800
},
{
"epoch": 1.8,
"grad_norm": 1.759525396989241,
"learning_rate": 1.6784200147193138e-05,
"loss": 0.1212,
"step": 12820
},
{
"epoch": 1.8,
"grad_norm": 1.8662789660260795,
"learning_rate": 1.6772839256566032e-05,
"loss": 0.1084,
"step": 12840
},
{
"epoch": 1.8,
"grad_norm": 1.1644106350747554,
"learning_rate": 1.6761462193546617e-05,
"loss": 0.108,
"step": 12860
},
{
"epoch": 1.81,
"grad_norm": 1.637080987799325,
"learning_rate": 1.6750068985305397e-05,
"loss": 0.1156,
"step": 12880
},
{
"epoch": 1.81,
"grad_norm": 1.6333523090713484,
"learning_rate": 1.6738659659051448e-05,
"loss": 0.1126,
"step": 12900
},
{
"epoch": 1.81,
"grad_norm": 1.4385397095037356,
"learning_rate": 1.6727234242032327e-05,
"loss": 0.1122,
"step": 12920
},
{
"epoch": 1.81,
"grad_norm": 1.3426431488637385,
"learning_rate": 1.671579276153402e-05,
"loss": 0.1098,
"step": 12940
},
{
"epoch": 1.82,
"grad_norm": 1.2419827370449787,
"learning_rate": 1.6704335244880882e-05,
"loss": 0.1098,
"step": 12960
},
{
"epoch": 1.82,
"grad_norm": 1.2398069591387826,
"learning_rate": 1.669286171943556e-05,
"loss": 0.1031,
"step": 12980
},
{
"epoch": 1.82,
"grad_norm": 1.3059608048082352,
"learning_rate": 1.6681372212598933e-05,
"loss": 0.1077,
"step": 13000
},
{
"epoch": 1.83,
"grad_norm": 6.89749875275012,
"learning_rate": 1.6669866751810043e-05,
"loss": 0.1116,
"step": 13020
},
{
"epoch": 1.83,
"grad_norm": 1.5961823695111663,
"learning_rate": 1.6658345364546046e-05,
"loss": 0.1131,
"step": 13040
},
{
"epoch": 1.83,
"grad_norm": 2.727450338273422,
"learning_rate": 1.664680807832212e-05,
"loss": 0.1112,
"step": 13060
},
{
"epoch": 1.83,
"grad_norm": 13.615360314696906,
"learning_rate": 1.6635254920691417e-05,
"loss": 0.1032,
"step": 13080
},
{
"epoch": 1.84,
"grad_norm": 2.1244072929008624,
"learning_rate": 1.6623685919244993e-05,
"loss": 0.1026,
"step": 13100
},
{
"epoch": 1.84,
"grad_norm": 1.283583261048826,
"learning_rate": 1.661210110161175e-05,
"loss": 0.0989,
"step": 13120
},
{
"epoch": 1.84,
"grad_norm": 1.3090789350951824,
"learning_rate": 1.660050049545834e-05,
"loss": 0.1024,
"step": 13140
},
{
"epoch": 1.84,
"grad_norm": 0.9759148548246103,
"learning_rate": 1.6588884128489144e-05,
"loss": 0.1098,
"step": 13160
},
{
"epoch": 1.85,
"grad_norm": 1.2587740048616956,
"learning_rate": 1.6577252028446175e-05,
"loss": 0.1067,
"step": 13180
},
{
"epoch": 1.85,
"grad_norm": 1.2714497316859061,
"learning_rate": 1.6565604223109014e-05,
"loss": 0.1134,
"step": 13200
},
{
"epoch": 1.85,
"grad_norm": 1.5669803471066202,
"learning_rate": 1.6553940740294755e-05,
"loss": 0.1139,
"step": 13220
},
{
"epoch": 1.86,
"grad_norm": 1.5888253214079329,
"learning_rate": 1.654226160785793e-05,
"loss": 0.1125,
"step": 13240
},
{
"epoch": 1.86,
"grad_norm": 1.498137907721368,
"learning_rate": 1.6530566853690448e-05,
"loss": 0.1105,
"step": 13260
},
{
"epoch": 1.86,
"grad_norm": 1.6902720677030239,
"learning_rate": 1.651885650572152e-05,
"loss": 0.1093,
"step": 13280
},
{
"epoch": 1.86,
"grad_norm": 2.347457496493911,
"learning_rate": 1.6507130591917603e-05,
"loss": 0.1098,
"step": 13300
},
{
"epoch": 1.87,
"grad_norm": 1.3678674369518569,
"learning_rate": 1.6495389140282335e-05,
"loss": 0.1098,
"step": 13320
},
{
"epoch": 1.87,
"grad_norm": 1.5629724169864214,
"learning_rate": 1.6483632178856438e-05,
"loss": 0.1073,
"step": 13340
},
{
"epoch": 1.87,
"grad_norm": 0.993333830150591,
"learning_rate": 1.64718597357177e-05,
"loss": 0.111,
"step": 13360
},
{
"epoch": 1.88,
"grad_norm": 1.4094609371806337,
"learning_rate": 1.646007183898086e-05,
"loss": 0.1077,
"step": 13380
},
{
"epoch": 1.88,
"grad_norm": 1.564626543622567,
"learning_rate": 1.644826851679759e-05,
"loss": 0.1123,
"step": 13400
},
{
"epoch": 1.88,
"grad_norm": 1.4612973020026034,
"learning_rate": 1.6436449797356374e-05,
"loss": 0.1092,
"step": 13420
},
{
"epoch": 1.88,
"grad_norm": 1.3376027041067962,
"learning_rate": 1.6424615708882483e-05,
"loss": 0.1067,
"step": 13440
},
{
"epoch": 1.89,
"grad_norm": 1.0775195647054387,
"learning_rate": 1.6412766279637882e-05,
"loss": 0.1069,
"step": 13460
},
{
"epoch": 1.89,
"grad_norm": 1.5157478518690226,
"learning_rate": 1.6400901537921182e-05,
"loss": 0.1057,
"step": 13480
},
{
"epoch": 1.89,
"grad_norm": 1.2971219666427078,
"learning_rate": 1.6389021512067564e-05,
"loss": 0.1152,
"step": 13500
},
{
"epoch": 1.9,
"grad_norm": 1.5813657514702586,
"learning_rate": 1.6377126230448694e-05,
"loss": 0.107,
"step": 13520
},
{
"epoch": 1.9,
"grad_norm": 1.7866841461740508,
"learning_rate": 1.6365215721472695e-05,
"loss": 0.1128,
"step": 13540
},
{
"epoch": 1.9,
"grad_norm": 1.315689661607094,
"learning_rate": 1.6353290013584034e-05,
"loss": 0.0961,
"step": 13560
},
{
"epoch": 1.9,
"grad_norm": 1.5657635747378045,
"learning_rate": 1.634134913526349e-05,
"loss": 0.1161,
"step": 13580
},
{
"epoch": 1.91,
"grad_norm": 1.9685502934967456,
"learning_rate": 1.6329393115028064e-05,
"loss": 0.1009,
"step": 13600
},
{
"epoch": 1.91,
"grad_norm": 2.0725566864544125,
"learning_rate": 1.6317421981430928e-05,
"loss": 0.1093,
"step": 13620
},
{
"epoch": 1.91,
"grad_norm": 1.7101568243828267,
"learning_rate": 1.6305435763061337e-05,
"loss": 0.1093,
"step": 13640
},
{
"epoch": 1.92,
"grad_norm": 1.4542610075915856,
"learning_rate": 1.6293434488544574e-05,
"loss": 0.1037,
"step": 13660
},
{
"epoch": 1.92,
"grad_norm": 1.371835926168549,
"learning_rate": 1.6281418186541885e-05,
"loss": 0.0995,
"step": 13680
},
{
"epoch": 1.92,
"grad_norm": 2.3100083820801953,
"learning_rate": 1.62693868857504e-05,
"loss": 0.1103,
"step": 13700
},
{
"epoch": 1.92,
"grad_norm": 1.5896808812652474,
"learning_rate": 1.6257340614903064e-05,
"loss": 0.1102,
"step": 13720
},
{
"epoch": 1.93,
"grad_norm": 1.417215951383714,
"learning_rate": 1.624527940276858e-05,
"loss": 0.1041,
"step": 13740
},
{
"epoch": 1.93,
"grad_norm": 1.7404045252992562,
"learning_rate": 1.623320327815133e-05,
"loss": 0.1096,
"step": 13760
},
{
"epoch": 1.93,
"grad_norm": 1.33620847218201,
"learning_rate": 1.6221112269891313e-05,
"loss": 0.114,
"step": 13780
},
{
"epoch": 1.93,
"grad_norm": 1.5758363350056759,
"learning_rate": 1.6209006406864073e-05,
"loss": 0.1111,
"step": 13800
},
{
"epoch": 1.94,
"grad_norm": 1.4315293751539684,
"learning_rate": 1.6196885717980623e-05,
"loss": 0.1102,
"step": 13820
},
{
"epoch": 1.94,
"grad_norm": 17.93867217696606,
"learning_rate": 1.6184750232187392e-05,
"loss": 0.1068,
"step": 13840
},
{
"epoch": 1.94,
"grad_norm": 1.4976416963716854,
"learning_rate": 1.617259997846614e-05,
"loss": 0.1119,
"step": 13860
},
{
"epoch": 1.95,
"grad_norm": 2.6760590569176235,
"learning_rate": 1.61604349858339e-05,
"loss": 0.1029,
"step": 13880
},
{
"epoch": 1.95,
"grad_norm": 1.268143867062791,
"learning_rate": 1.6148255283342905e-05,
"loss": 0.1075,
"step": 13900
},
{
"epoch": 1.95,
"grad_norm": 1.2654550226298094,
"learning_rate": 1.6136060900080507e-05,
"loss": 0.1139,
"step": 13920
},
{
"epoch": 1.95,
"grad_norm": 1.7276044782520006,
"learning_rate": 1.6123851865169128e-05,
"loss": 0.117,
"step": 13940
},
{
"epoch": 1.96,
"grad_norm": 1.5419376839589751,
"learning_rate": 1.6111628207766184e-05,
"loss": 0.1116,
"step": 13960
},
{
"epoch": 1.96,
"grad_norm": 1.5839809883354286,
"learning_rate": 1.6099389957064008e-05,
"loss": 0.1038,
"step": 13980
},
{
"epoch": 1.96,
"grad_norm": 1.6072993550474162,
"learning_rate": 1.608713714228978e-05,
"loss": 0.1086,
"step": 14000
},
{
"epoch": 1.97,
"grad_norm": 1.6826229969883024,
"learning_rate": 1.6074869792705463e-05,
"loss": 0.1143,
"step": 14020
},
{
"epoch": 1.97,
"grad_norm": 1.751260507839353,
"learning_rate": 1.6062587937607744e-05,
"loss": 0.1071,
"step": 14040
},
{
"epoch": 1.97,
"grad_norm": 1.787402653210854,
"learning_rate": 1.6050291606327935e-05,
"loss": 0.1087,
"step": 14060
},
{
"epoch": 1.97,
"grad_norm": 1.4597084464985797,
"learning_rate": 1.603798082823193e-05,
"loss": 0.1073,
"step": 14080
},
{
"epoch": 1.98,
"grad_norm": 1.9309402908821143,
"learning_rate": 1.602565563272012e-05,
"loss": 0.1109,
"step": 14100
},
{
"epoch": 1.98,
"grad_norm": 1.1027110719938062,
"learning_rate": 1.6013316049227327e-05,
"loss": 0.1015,
"step": 14120
},
{
"epoch": 1.98,
"grad_norm": 1.5121790458693565,
"learning_rate": 1.600096210722274e-05,
"loss": 0.1094,
"step": 14140
},
{
"epoch": 1.99,
"grad_norm": 1.1973659852920389,
"learning_rate": 1.5988593836209834e-05,
"loss": 0.1118,
"step": 14160
},
{
"epoch": 1.99,
"grad_norm": 1.258094045274343,
"learning_rate": 1.5976211265726305e-05,
"loss": 0.1068,
"step": 14180
},
{
"epoch": 1.99,
"grad_norm": 1.1547212546735568,
"learning_rate": 1.5963814425343994e-05,
"loss": 0.1,
"step": 14200
},
{
"epoch": 1.99,
"grad_norm": 1.391967382701112,
"learning_rate": 1.5951403344668836e-05,
"loss": 0.1072,
"step": 14220
},
{
"epoch": 2.0,
"grad_norm": 1.2528703159647003,
"learning_rate": 1.593897805334076e-05,
"loss": 0.1021,
"step": 14240
},
{
"epoch": 2.0,
"grad_norm": 2.0691927202755567,
"learning_rate": 1.5926538581033636e-05,
"loss": 0.1033,
"step": 14260
}
],
"logging_steps": 20,
"max_steps": 42798,
"num_input_tokens_seen": 0,
"num_train_epochs": 6,
"save_steps": 500,
"total_flos": 0.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}