m2m100_418M_swa_en_rel / trainer_state.json
Davlan's picture
add MT model
62b6276
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"global_step": 260853,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.990416058086355e-05,
"loss": 2.1636,
"step": 500
},
{
"epoch": 0.01,
"learning_rate": 4.9808321161727104e-05,
"loss": 1.994,
"step": 1000
},
{
"epoch": 0.02,
"learning_rate": 4.971248174259066e-05,
"loss": 1.907,
"step": 1500
},
{
"epoch": 0.02,
"learning_rate": 4.961664232345421e-05,
"loss": 1.8301,
"step": 2000
},
{
"epoch": 0.03,
"learning_rate": 4.952080290431776e-05,
"loss": 1.7698,
"step": 2500
},
{
"epoch": 0.03,
"learning_rate": 4.942496348518131e-05,
"loss": 1.7158,
"step": 3000
},
{
"epoch": 0.04,
"learning_rate": 4.932912406604486e-05,
"loss": 1.6953,
"step": 3500
},
{
"epoch": 0.05,
"learning_rate": 4.9233284646908415e-05,
"loss": 1.6492,
"step": 4000
},
{
"epoch": 0.05,
"learning_rate": 4.913744522777197e-05,
"loss": 1.6356,
"step": 4500
},
{
"epoch": 0.06,
"learning_rate": 4.9041605808635516e-05,
"loss": 1.6396,
"step": 5000
},
{
"epoch": 0.06,
"learning_rate": 4.894576638949907e-05,
"loss": 1.6158,
"step": 5500
},
{
"epoch": 0.07,
"learning_rate": 4.884992697036262e-05,
"loss": 1.5995,
"step": 6000
},
{
"epoch": 0.07,
"learning_rate": 4.875408755122617e-05,
"loss": 1.5564,
"step": 6500
},
{
"epoch": 0.08,
"learning_rate": 4.8658248132089725e-05,
"loss": 1.5438,
"step": 7000
},
{
"epoch": 0.09,
"learning_rate": 4.856240871295327e-05,
"loss": 1.5618,
"step": 7500
},
{
"epoch": 0.09,
"learning_rate": 4.846656929381683e-05,
"loss": 1.5249,
"step": 8000
},
{
"epoch": 0.1,
"learning_rate": 4.8370729874680374e-05,
"loss": 1.5021,
"step": 8500
},
{
"epoch": 0.1,
"learning_rate": 4.827489045554393e-05,
"loss": 1.4965,
"step": 9000
},
{
"epoch": 0.11,
"learning_rate": 4.817905103640748e-05,
"loss": 1.482,
"step": 9500
},
{
"epoch": 0.12,
"learning_rate": 4.8083211617271036e-05,
"loss": 1.478,
"step": 10000
},
{
"epoch": 0.12,
"learning_rate": 4.798737219813458e-05,
"loss": 1.4683,
"step": 10500
},
{
"epoch": 0.13,
"learning_rate": 4.789153277899813e-05,
"loss": 1.4413,
"step": 11000
},
{
"epoch": 0.13,
"learning_rate": 4.7795693359861685e-05,
"loss": 1.4603,
"step": 11500
},
{
"epoch": 0.14,
"learning_rate": 4.769985394072524e-05,
"loss": 1.4551,
"step": 12000
},
{
"epoch": 0.14,
"learning_rate": 4.760401452158879e-05,
"loss": 1.443,
"step": 12500
},
{
"epoch": 0.15,
"learning_rate": 4.750817510245234e-05,
"loss": 1.4114,
"step": 13000
},
{
"epoch": 0.16,
"learning_rate": 4.7412335683315894e-05,
"loss": 1.4287,
"step": 13500
},
{
"epoch": 0.16,
"learning_rate": 4.731649626417944e-05,
"loss": 1.4237,
"step": 14000
},
{
"epoch": 0.17,
"learning_rate": 4.7220656845042995e-05,
"loss": 1.4067,
"step": 14500
},
{
"epoch": 0.17,
"learning_rate": 4.712481742590655e-05,
"loss": 1.4095,
"step": 15000
},
{
"epoch": 0.18,
"learning_rate": 4.70289780067701e-05,
"loss": 1.3931,
"step": 15500
},
{
"epoch": 0.18,
"learning_rate": 4.693313858763365e-05,
"loss": 1.3703,
"step": 16000
},
{
"epoch": 0.19,
"learning_rate": 4.68372991684972e-05,
"loss": 1.3877,
"step": 16500
},
{
"epoch": 0.2,
"learning_rate": 4.674145974936075e-05,
"loss": 1.3768,
"step": 17000
},
{
"epoch": 0.2,
"learning_rate": 4.6645620330224306e-05,
"loss": 1.3745,
"step": 17500
},
{
"epoch": 0.21,
"learning_rate": 4.6549780911087854e-05,
"loss": 1.3683,
"step": 18000
},
{
"epoch": 0.21,
"learning_rate": 4.645394149195141e-05,
"loss": 1.3696,
"step": 18500
},
{
"epoch": 0.22,
"learning_rate": 4.6358102072814955e-05,
"loss": 1.3424,
"step": 19000
},
{
"epoch": 0.22,
"learning_rate": 4.626226265367851e-05,
"loss": 1.3405,
"step": 19500
},
{
"epoch": 0.23,
"learning_rate": 4.616642323454206e-05,
"loss": 1.3524,
"step": 20000
},
{
"epoch": 0.24,
"learning_rate": 4.607058381540562e-05,
"loss": 1.3253,
"step": 20500
},
{
"epoch": 0.24,
"learning_rate": 4.5974744396269164e-05,
"loss": 1.323,
"step": 21000
},
{
"epoch": 0.25,
"learning_rate": 4.587890497713272e-05,
"loss": 1.3265,
"step": 21500
},
{
"epoch": 0.25,
"learning_rate": 4.5783065557996266e-05,
"loss": 1.3285,
"step": 22000
},
{
"epoch": 0.26,
"learning_rate": 4.568722613885982e-05,
"loss": 1.3064,
"step": 22500
},
{
"epoch": 0.26,
"learning_rate": 4.5591386719723374e-05,
"loss": 1.3208,
"step": 23000
},
{
"epoch": 0.27,
"learning_rate": 4.549554730058692e-05,
"loss": 1.3224,
"step": 23500
},
{
"epoch": 0.28,
"learning_rate": 4.5399707881450475e-05,
"loss": 1.3329,
"step": 24000
},
{
"epoch": 0.28,
"learning_rate": 4.530386846231402e-05,
"loss": 1.318,
"step": 24500
},
{
"epoch": 0.29,
"learning_rate": 4.520802904317758e-05,
"loss": 1.3029,
"step": 25000
},
{
"epoch": 0.29,
"learning_rate": 4.511218962404113e-05,
"loss": 1.3139,
"step": 25500
},
{
"epoch": 0.3,
"learning_rate": 4.501635020490468e-05,
"loss": 1.3033,
"step": 26000
},
{
"epoch": 0.3,
"learning_rate": 4.492051078576823e-05,
"loss": 1.3046,
"step": 26500
},
{
"epoch": 0.31,
"learning_rate": 4.482467136663178e-05,
"loss": 1.2975,
"step": 27000
},
{
"epoch": 0.32,
"learning_rate": 4.472883194749533e-05,
"loss": 1.2869,
"step": 27500
},
{
"epoch": 0.32,
"learning_rate": 4.463299252835889e-05,
"loss": 1.2797,
"step": 28000
},
{
"epoch": 0.33,
"learning_rate": 4.453715310922244e-05,
"loss": 1.2803,
"step": 28500
},
{
"epoch": 0.33,
"learning_rate": 4.444131369008599e-05,
"loss": 1.2613,
"step": 29000
},
{
"epoch": 0.34,
"learning_rate": 4.4345474270949536e-05,
"loss": 1.2781,
"step": 29500
},
{
"epoch": 0.35,
"learning_rate": 4.424963485181309e-05,
"loss": 1.2875,
"step": 30000
},
{
"epoch": 0.35,
"learning_rate": 4.4153795432676644e-05,
"loss": 1.2775,
"step": 30500
},
{
"epoch": 0.36,
"learning_rate": 4.40579560135402e-05,
"loss": 1.272,
"step": 31000
},
{
"epoch": 0.36,
"learning_rate": 4.3962116594403745e-05,
"loss": 1.265,
"step": 31500
},
{
"epoch": 0.37,
"learning_rate": 4.38662771752673e-05,
"loss": 1.2522,
"step": 32000
},
{
"epoch": 0.37,
"learning_rate": 4.377043775613085e-05,
"loss": 1.262,
"step": 32500
},
{
"epoch": 0.38,
"learning_rate": 4.36745983369944e-05,
"loss": 1.2353,
"step": 33000
},
{
"epoch": 0.39,
"learning_rate": 4.3578758917857955e-05,
"loss": 1.2425,
"step": 33500
},
{
"epoch": 0.39,
"learning_rate": 4.34829194987215e-05,
"loss": 1.2536,
"step": 34000
},
{
"epoch": 0.4,
"learning_rate": 4.3387080079585056e-05,
"loss": 1.2463,
"step": 34500
},
{
"epoch": 0.4,
"learning_rate": 4.3291240660448603e-05,
"loss": 1.2224,
"step": 35000
},
{
"epoch": 0.41,
"learning_rate": 4.3195401241312164e-05,
"loss": 1.2313,
"step": 35500
},
{
"epoch": 0.41,
"learning_rate": 4.309956182217571e-05,
"loss": 1.2473,
"step": 36000
},
{
"epoch": 0.42,
"learning_rate": 4.300372240303926e-05,
"loss": 1.2193,
"step": 36500
},
{
"epoch": 0.43,
"learning_rate": 4.290788298390281e-05,
"loss": 1.2234,
"step": 37000
},
{
"epoch": 0.43,
"learning_rate": 4.281204356476636e-05,
"loss": 1.2365,
"step": 37500
},
{
"epoch": 0.44,
"learning_rate": 4.271620414562992e-05,
"loss": 1.2107,
"step": 38000
},
{
"epoch": 0.44,
"learning_rate": 4.262036472649347e-05,
"loss": 1.219,
"step": 38500
},
{
"epoch": 0.45,
"learning_rate": 4.252452530735702e-05,
"loss": 1.232,
"step": 39000
},
{
"epoch": 0.45,
"learning_rate": 4.242868588822057e-05,
"loss": 1.2158,
"step": 39500
},
{
"epoch": 0.46,
"learning_rate": 4.233284646908412e-05,
"loss": 1.2113,
"step": 40000
},
{
"epoch": 0.47,
"learning_rate": 4.223700704994768e-05,
"loss": 1.2027,
"step": 40500
},
{
"epoch": 0.47,
"learning_rate": 4.2141167630811225e-05,
"loss": 1.2351,
"step": 41000
},
{
"epoch": 0.48,
"learning_rate": 4.204532821167478e-05,
"loss": 1.2031,
"step": 41500
},
{
"epoch": 0.48,
"learning_rate": 4.1949488792538326e-05,
"loss": 1.2189,
"step": 42000
},
{
"epoch": 0.49,
"learning_rate": 4.185364937340188e-05,
"loss": 1.2178,
"step": 42500
},
{
"epoch": 0.49,
"learning_rate": 4.1757809954265434e-05,
"loss": 1.2127,
"step": 43000
},
{
"epoch": 0.5,
"learning_rate": 4.166197053512898e-05,
"loss": 1.2245,
"step": 43500
},
{
"epoch": 0.51,
"learning_rate": 4.1566131115992536e-05,
"loss": 1.2316,
"step": 44000
},
{
"epoch": 0.51,
"learning_rate": 4.147029169685608e-05,
"loss": 1.2077,
"step": 44500
},
{
"epoch": 0.52,
"learning_rate": 4.137445227771964e-05,
"loss": 1.1855,
"step": 45000
},
{
"epoch": 0.52,
"learning_rate": 4.1278612858583184e-05,
"loss": 1.2007,
"step": 45500
},
{
"epoch": 0.53,
"learning_rate": 4.1182773439446745e-05,
"loss": 1.2152,
"step": 46000
},
{
"epoch": 0.53,
"learning_rate": 4.108693402031029e-05,
"loss": 1.2201,
"step": 46500
},
{
"epoch": 0.54,
"learning_rate": 4.099109460117384e-05,
"loss": 1.2002,
"step": 47000
},
{
"epoch": 0.55,
"learning_rate": 4.0895255182037394e-05,
"loss": 1.2046,
"step": 47500
},
{
"epoch": 0.55,
"learning_rate": 4.079941576290094e-05,
"loss": 1.1818,
"step": 48000
},
{
"epoch": 0.56,
"learning_rate": 4.07035763437645e-05,
"loss": 1.1897,
"step": 48500
},
{
"epoch": 0.56,
"learning_rate": 4.060773692462805e-05,
"loss": 1.1969,
"step": 49000
},
{
"epoch": 0.57,
"learning_rate": 4.05118975054916e-05,
"loss": 1.1983,
"step": 49500
},
{
"epoch": 0.58,
"learning_rate": 4.041605808635515e-05,
"loss": 1.192,
"step": 50000
},
{
"epoch": 0.58,
"learning_rate": 4.03202186672187e-05,
"loss": 1.1977,
"step": 50500
},
{
"epoch": 0.59,
"learning_rate": 4.022437924808226e-05,
"loss": 1.1815,
"step": 51000
},
{
"epoch": 0.59,
"learning_rate": 4.0128539828945806e-05,
"loss": 1.1385,
"step": 51500
},
{
"epoch": 0.6,
"learning_rate": 4.003270040980936e-05,
"loss": 1.1852,
"step": 52000
},
{
"epoch": 0.6,
"learning_rate": 3.993686099067291e-05,
"loss": 1.1639,
"step": 52500
},
{
"epoch": 0.61,
"learning_rate": 3.984102157153646e-05,
"loss": 1.1693,
"step": 53000
},
{
"epoch": 0.62,
"learning_rate": 3.9745182152400015e-05,
"loss": 1.1489,
"step": 53500
},
{
"epoch": 0.62,
"learning_rate": 3.964934273326356e-05,
"loss": 1.1661,
"step": 54000
},
{
"epoch": 0.63,
"learning_rate": 3.955350331412712e-05,
"loss": 1.1732,
"step": 54500
},
{
"epoch": 0.63,
"learning_rate": 3.9457663894990664e-05,
"loss": 1.1839,
"step": 55000
},
{
"epoch": 0.64,
"learning_rate": 3.936182447585422e-05,
"loss": 1.1767,
"step": 55500
},
{
"epoch": 0.64,
"learning_rate": 3.926598505671777e-05,
"loss": 1.17,
"step": 56000
},
{
"epoch": 0.65,
"learning_rate": 3.9170145637581326e-05,
"loss": 1.1701,
"step": 56500
},
{
"epoch": 0.66,
"learning_rate": 3.9074306218444873e-05,
"loss": 1.1392,
"step": 57000
},
{
"epoch": 0.66,
"learning_rate": 3.897846679930843e-05,
"loss": 1.1493,
"step": 57500
},
{
"epoch": 0.67,
"learning_rate": 3.8882627380171975e-05,
"loss": 1.1413,
"step": 58000
},
{
"epoch": 0.67,
"learning_rate": 3.878678796103553e-05,
"loss": 1.1587,
"step": 58500
},
{
"epoch": 0.68,
"learning_rate": 3.869094854189908e-05,
"loss": 1.1577,
"step": 59000
},
{
"epoch": 0.68,
"learning_rate": 3.859510912276263e-05,
"loss": 1.1495,
"step": 59500
},
{
"epoch": 0.69,
"learning_rate": 3.8499269703626184e-05,
"loss": 1.1534,
"step": 60000
},
{
"epoch": 0.7,
"learning_rate": 3.840343028448973e-05,
"loss": 1.165,
"step": 60500
},
{
"epoch": 0.7,
"learning_rate": 3.8307590865353286e-05,
"loss": 1.1444,
"step": 61000
},
{
"epoch": 0.71,
"learning_rate": 3.821175144621684e-05,
"loss": 1.1464,
"step": 61500
},
{
"epoch": 0.71,
"learning_rate": 3.811591202708039e-05,
"loss": 1.1475,
"step": 62000
},
{
"epoch": 0.72,
"learning_rate": 3.802007260794394e-05,
"loss": 1.14,
"step": 62500
},
{
"epoch": 0.72,
"learning_rate": 3.792423318880749e-05,
"loss": 1.1265,
"step": 63000
},
{
"epoch": 0.73,
"learning_rate": 3.782839376967104e-05,
"loss": 1.1388,
"step": 63500
},
{
"epoch": 0.74,
"learning_rate": 3.7732554350534596e-05,
"loss": 1.1373,
"step": 64000
},
{
"epoch": 0.74,
"learning_rate": 3.763671493139815e-05,
"loss": 1.1309,
"step": 64500
},
{
"epoch": 0.75,
"learning_rate": 3.75408755122617e-05,
"loss": 1.1304,
"step": 65000
},
{
"epoch": 0.75,
"learning_rate": 3.7445036093125245e-05,
"loss": 1.1173,
"step": 65500
},
{
"epoch": 0.76,
"learning_rate": 3.73491966739888e-05,
"loss": 1.12,
"step": 66000
},
{
"epoch": 0.76,
"learning_rate": 3.725335725485235e-05,
"loss": 1.1116,
"step": 66500
},
{
"epoch": 0.77,
"learning_rate": 3.715751783571591e-05,
"loss": 1.149,
"step": 67000
},
{
"epoch": 0.78,
"learning_rate": 3.7061678416579454e-05,
"loss": 1.1255,
"step": 67500
},
{
"epoch": 0.78,
"learning_rate": 3.696583899744301e-05,
"loss": 1.1429,
"step": 68000
},
{
"epoch": 0.79,
"learning_rate": 3.6869999578306556e-05,
"loss": 1.1217,
"step": 68500
},
{
"epoch": 0.79,
"learning_rate": 3.677416015917011e-05,
"loss": 1.1174,
"step": 69000
},
{
"epoch": 0.8,
"learning_rate": 3.6678320740033664e-05,
"loss": 1.113,
"step": 69500
},
{
"epoch": 0.81,
"learning_rate": 3.658248132089721e-05,
"loss": 1.1208,
"step": 70000
},
{
"epoch": 0.81,
"learning_rate": 3.6486641901760765e-05,
"loss": 1.1174,
"step": 70500
},
{
"epoch": 0.82,
"learning_rate": 3.639080248262431e-05,
"loss": 1.1201,
"step": 71000
},
{
"epoch": 0.82,
"learning_rate": 3.6294963063487867e-05,
"loss": 1.1266,
"step": 71500
},
{
"epoch": 0.83,
"learning_rate": 3.619912364435142e-05,
"loss": 1.1201,
"step": 72000
},
{
"epoch": 0.83,
"learning_rate": 3.610328422521497e-05,
"loss": 1.1003,
"step": 72500
},
{
"epoch": 0.84,
"learning_rate": 3.600744480607852e-05,
"loss": 1.1004,
"step": 73000
},
{
"epoch": 0.85,
"learning_rate": 3.591160538694207e-05,
"loss": 1.1038,
"step": 73500
},
{
"epoch": 0.85,
"learning_rate": 3.581576596780562e-05,
"loss": 1.1264,
"step": 74000
},
{
"epoch": 0.86,
"learning_rate": 3.571992654866918e-05,
"loss": 1.0977,
"step": 74500
},
{
"epoch": 0.86,
"learning_rate": 3.562408712953273e-05,
"loss": 1.1079,
"step": 75000
},
{
"epoch": 0.87,
"learning_rate": 3.552824771039628e-05,
"loss": 1.1059,
"step": 75500
},
{
"epoch": 0.87,
"learning_rate": 3.5432408291259826e-05,
"loss": 1.1193,
"step": 76000
},
{
"epoch": 0.88,
"learning_rate": 3.533656887212338e-05,
"loss": 1.1115,
"step": 76500
},
{
"epoch": 0.89,
"learning_rate": 3.5240729452986934e-05,
"loss": 1.1191,
"step": 77000
},
{
"epoch": 0.89,
"learning_rate": 3.514489003385049e-05,
"loss": 1.1028,
"step": 77500
},
{
"epoch": 0.9,
"learning_rate": 3.5049050614714035e-05,
"loss": 1.0868,
"step": 78000
},
{
"epoch": 0.9,
"learning_rate": 3.495321119557759e-05,
"loss": 1.1205,
"step": 78500
},
{
"epoch": 0.91,
"learning_rate": 3.485737177644114e-05,
"loss": 1.0953,
"step": 79000
},
{
"epoch": 0.91,
"learning_rate": 3.476153235730469e-05,
"loss": 1.0871,
"step": 79500
},
{
"epoch": 0.92,
"learning_rate": 3.4665692938168245e-05,
"loss": 1.0888,
"step": 80000
},
{
"epoch": 0.93,
"learning_rate": 3.456985351903179e-05,
"loss": 1.0983,
"step": 80500
},
{
"epoch": 0.93,
"learning_rate": 3.4474014099895346e-05,
"loss": 1.1028,
"step": 81000
},
{
"epoch": 0.94,
"learning_rate": 3.4378174680758893e-05,
"loss": 1.0996,
"step": 81500
},
{
"epoch": 0.94,
"learning_rate": 3.428233526162245e-05,
"loss": 1.0894,
"step": 82000
},
{
"epoch": 0.95,
"learning_rate": 3.4186495842486e-05,
"loss": 1.099,
"step": 82500
},
{
"epoch": 0.95,
"learning_rate": 3.409065642334955e-05,
"loss": 1.0693,
"step": 83000
},
{
"epoch": 0.96,
"learning_rate": 3.39948170042131e-05,
"loss": 1.1003,
"step": 83500
},
{
"epoch": 0.97,
"learning_rate": 3.389897758507665e-05,
"loss": 1.0709,
"step": 84000
},
{
"epoch": 0.97,
"learning_rate": 3.3803138165940204e-05,
"loss": 1.098,
"step": 84500
},
{
"epoch": 0.98,
"learning_rate": 3.370729874680376e-05,
"loss": 1.0889,
"step": 85000
},
{
"epoch": 0.98,
"learning_rate": 3.361145932766731e-05,
"loss": 1.1002,
"step": 85500
},
{
"epoch": 0.99,
"learning_rate": 3.351561990853086e-05,
"loss": 1.0914,
"step": 86000
},
{
"epoch": 0.99,
"learning_rate": 3.341978048939441e-05,
"loss": 1.0916,
"step": 86500
},
{
"epoch": 1.0,
"learning_rate": 3.332394107025796e-05,
"loss": 1.0824,
"step": 87000
},
{
"epoch": 1.01,
"learning_rate": 3.3228101651121515e-05,
"loss": 0.9678,
"step": 87500
},
{
"epoch": 1.01,
"learning_rate": 3.313226223198507e-05,
"loss": 0.9703,
"step": 88000
},
{
"epoch": 1.02,
"learning_rate": 3.3036422812848616e-05,
"loss": 0.9833,
"step": 88500
},
{
"epoch": 1.02,
"learning_rate": 3.294058339371217e-05,
"loss": 0.9731,
"step": 89000
},
{
"epoch": 1.03,
"learning_rate": 3.284474397457572e-05,
"loss": 0.9738,
"step": 89500
},
{
"epoch": 1.04,
"learning_rate": 3.274890455543927e-05,
"loss": 0.9739,
"step": 90000
},
{
"epoch": 1.04,
"learning_rate": 3.2653065136302826e-05,
"loss": 0.9543,
"step": 90500
},
{
"epoch": 1.05,
"learning_rate": 3.255722571716637e-05,
"loss": 0.9771,
"step": 91000
},
{
"epoch": 1.05,
"learning_rate": 3.246138629802993e-05,
"loss": 0.9846,
"step": 91500
},
{
"epoch": 1.06,
"learning_rate": 3.2365546878893474e-05,
"loss": 0.9806,
"step": 92000
},
{
"epoch": 1.06,
"learning_rate": 3.226970745975703e-05,
"loss": 0.9792,
"step": 92500
},
{
"epoch": 1.07,
"learning_rate": 3.217386804062058e-05,
"loss": 0.9908,
"step": 93000
},
{
"epoch": 1.08,
"learning_rate": 3.207802862148413e-05,
"loss": 0.9726,
"step": 93500
},
{
"epoch": 1.08,
"learning_rate": 3.1982189202347684e-05,
"loss": 0.9701,
"step": 94000
},
{
"epoch": 1.09,
"learning_rate": 3.188634978321123e-05,
"loss": 0.9697,
"step": 94500
},
{
"epoch": 1.09,
"learning_rate": 3.1790510364074785e-05,
"loss": 0.9535,
"step": 95000
},
{
"epoch": 1.1,
"learning_rate": 3.169467094493834e-05,
"loss": 0.9686,
"step": 95500
},
{
"epoch": 1.1,
"learning_rate": 3.159883152580189e-05,
"loss": 0.9778,
"step": 96000
},
{
"epoch": 1.11,
"learning_rate": 3.150299210666544e-05,
"loss": 0.9699,
"step": 96500
},
{
"epoch": 1.12,
"learning_rate": 3.1407152687528995e-05,
"loss": 0.9701,
"step": 97000
},
{
"epoch": 1.12,
"learning_rate": 3.131131326839254e-05,
"loss": 0.9829,
"step": 97500
},
{
"epoch": 1.13,
"learning_rate": 3.1215473849256096e-05,
"loss": 0.9739,
"step": 98000
},
{
"epoch": 1.13,
"learning_rate": 3.111963443011965e-05,
"loss": 0.9735,
"step": 98500
},
{
"epoch": 1.14,
"learning_rate": 3.10237950109832e-05,
"loss": 0.9637,
"step": 99000
},
{
"epoch": 1.14,
"learning_rate": 3.092795559184675e-05,
"loss": 0.9675,
"step": 99500
},
{
"epoch": 1.15,
"learning_rate": 3.08321161727103e-05,
"loss": 0.9702,
"step": 100000
},
{
"epoch": 1.16,
"learning_rate": 3.073627675357385e-05,
"loss": 0.9812,
"step": 100500
},
{
"epoch": 1.16,
"learning_rate": 3.064043733443741e-05,
"loss": 0.9677,
"step": 101000
},
{
"epoch": 1.17,
"learning_rate": 3.0544597915300954e-05,
"loss": 0.9586,
"step": 101500
},
{
"epoch": 1.17,
"learning_rate": 3.0448758496164508e-05,
"loss": 0.9602,
"step": 102000
},
{
"epoch": 1.18,
"learning_rate": 3.035291907702806e-05,
"loss": 0.96,
"step": 102500
},
{
"epoch": 1.18,
"learning_rate": 3.0257079657891613e-05,
"loss": 0.9658,
"step": 103000
},
{
"epoch": 1.19,
"learning_rate": 3.016124023875516e-05,
"loss": 0.9756,
"step": 103500
},
{
"epoch": 1.2,
"learning_rate": 3.0065400819618718e-05,
"loss": 0.986,
"step": 104000
},
{
"epoch": 1.2,
"learning_rate": 2.9969561400482265e-05,
"loss": 0.9792,
"step": 104500
},
{
"epoch": 1.21,
"learning_rate": 2.9873721981345816e-05,
"loss": 0.9735,
"step": 105000
},
{
"epoch": 1.21,
"learning_rate": 2.977788256220937e-05,
"loss": 0.9672,
"step": 105500
},
{
"epoch": 1.22,
"learning_rate": 2.9682043143072917e-05,
"loss": 0.9558,
"step": 106000
},
{
"epoch": 1.22,
"learning_rate": 2.9586203723936474e-05,
"loss": 0.9605,
"step": 106500
},
{
"epoch": 1.23,
"learning_rate": 2.949036430480002e-05,
"loss": 0.9929,
"step": 107000
},
{
"epoch": 1.24,
"learning_rate": 2.9394524885663576e-05,
"loss": 0.9562,
"step": 107500
},
{
"epoch": 1.24,
"learning_rate": 2.9298685466527126e-05,
"loss": 0.9618,
"step": 108000
},
{
"epoch": 1.25,
"learning_rate": 2.9202846047390674e-05,
"loss": 0.9644,
"step": 108500
},
{
"epoch": 1.25,
"learning_rate": 2.910700662825423e-05,
"loss": 0.9558,
"step": 109000
},
{
"epoch": 1.26,
"learning_rate": 2.901116720911778e-05,
"loss": 0.9582,
"step": 109500
},
{
"epoch": 1.27,
"learning_rate": 2.8915327789981332e-05,
"loss": 0.9661,
"step": 110000
},
{
"epoch": 1.27,
"learning_rate": 2.8819488370844883e-05,
"loss": 0.9621,
"step": 110500
},
{
"epoch": 1.28,
"learning_rate": 2.8723648951708437e-05,
"loss": 0.9647,
"step": 111000
},
{
"epoch": 1.28,
"learning_rate": 2.8627809532571988e-05,
"loss": 0.9392,
"step": 111500
},
{
"epoch": 1.29,
"learning_rate": 2.8531970113435535e-05,
"loss": 0.9643,
"step": 112000
},
{
"epoch": 1.29,
"learning_rate": 2.843613069429909e-05,
"loss": 0.9783,
"step": 112500
},
{
"epoch": 1.3,
"learning_rate": 2.834029127516264e-05,
"loss": 0.9703,
"step": 113000
},
{
"epoch": 1.31,
"learning_rate": 2.8244451856026194e-05,
"loss": 0.9647,
"step": 113500
},
{
"epoch": 1.31,
"learning_rate": 2.8148612436889745e-05,
"loss": 0.9492,
"step": 114000
},
{
"epoch": 1.32,
"learning_rate": 2.80527730177533e-05,
"loss": 0.9603,
"step": 114500
},
{
"epoch": 1.32,
"learning_rate": 2.7956933598616846e-05,
"loss": 0.9602,
"step": 115000
},
{
"epoch": 1.33,
"learning_rate": 2.7861094179480397e-05,
"loss": 0.9663,
"step": 115500
},
{
"epoch": 1.33,
"learning_rate": 2.776525476034395e-05,
"loss": 0.9556,
"step": 116000
},
{
"epoch": 1.34,
"learning_rate": 2.76694153412075e-05,
"loss": 0.9564,
"step": 116500
},
{
"epoch": 1.35,
"learning_rate": 2.7573575922071055e-05,
"loss": 0.9301,
"step": 117000
},
{
"epoch": 1.35,
"learning_rate": 2.7477736502934603e-05,
"loss": 0.9284,
"step": 117500
},
{
"epoch": 1.36,
"learning_rate": 2.7381897083798157e-05,
"loss": 0.9444,
"step": 118000
},
{
"epoch": 1.36,
"learning_rate": 2.7286057664661707e-05,
"loss": 0.9767,
"step": 118500
},
{
"epoch": 1.37,
"learning_rate": 2.7190218245525258e-05,
"loss": 0.9633,
"step": 119000
},
{
"epoch": 1.37,
"learning_rate": 2.7094378826388812e-05,
"loss": 0.9468,
"step": 119500
},
{
"epoch": 1.38,
"learning_rate": 2.699853940725236e-05,
"loss": 0.9455,
"step": 120000
},
{
"epoch": 1.39,
"learning_rate": 2.6902699988115913e-05,
"loss": 0.9481,
"step": 120500
},
{
"epoch": 1.39,
"learning_rate": 2.6806860568979464e-05,
"loss": 0.9511,
"step": 121000
},
{
"epoch": 1.4,
"learning_rate": 2.6711021149843018e-05,
"loss": 0.9552,
"step": 121500
},
{
"epoch": 1.4,
"learning_rate": 2.661518173070657e-05,
"loss": 0.9626,
"step": 122000
},
{
"epoch": 1.41,
"learning_rate": 2.6519342311570116e-05,
"loss": 0.9528,
"step": 122500
},
{
"epoch": 1.41,
"learning_rate": 2.642350289243367e-05,
"loss": 0.9672,
"step": 123000
},
{
"epoch": 1.42,
"learning_rate": 2.632766347329722e-05,
"loss": 0.9479,
"step": 123500
},
{
"epoch": 1.43,
"learning_rate": 2.6231824054160775e-05,
"loss": 0.9369,
"step": 124000
},
{
"epoch": 1.43,
"learning_rate": 2.6135984635024326e-05,
"loss": 0.9545,
"step": 124500
},
{
"epoch": 1.44,
"learning_rate": 2.604014521588788e-05,
"loss": 0.9515,
"step": 125000
},
{
"epoch": 1.44,
"learning_rate": 2.5944305796751427e-05,
"loss": 0.9348,
"step": 125500
},
{
"epoch": 1.45,
"learning_rate": 2.5848466377614978e-05,
"loss": 0.9501,
"step": 126000
},
{
"epoch": 1.45,
"learning_rate": 2.575262695847853e-05,
"loss": 0.9564,
"step": 126500
},
{
"epoch": 1.46,
"learning_rate": 2.5656787539342082e-05,
"loss": 0.9367,
"step": 127000
},
{
"epoch": 1.47,
"learning_rate": 2.5560948120205636e-05,
"loss": 0.9495,
"step": 127500
},
{
"epoch": 1.47,
"learning_rate": 2.5465108701069184e-05,
"loss": 0.9328,
"step": 128000
},
{
"epoch": 1.48,
"learning_rate": 2.5369269281932738e-05,
"loss": 0.9475,
"step": 128500
},
{
"epoch": 1.48,
"learning_rate": 2.5273429862796288e-05,
"loss": 0.9601,
"step": 129000
},
{
"epoch": 1.49,
"learning_rate": 2.517759044365984e-05,
"loss": 0.9438,
"step": 129500
},
{
"epoch": 1.5,
"learning_rate": 2.5081751024523393e-05,
"loss": 0.9473,
"step": 130000
},
{
"epoch": 1.5,
"learning_rate": 2.4985911605386944e-05,
"loss": 0.9677,
"step": 130500
},
{
"epoch": 1.51,
"learning_rate": 2.4890072186250494e-05,
"loss": 0.9581,
"step": 131000
},
{
"epoch": 1.51,
"learning_rate": 2.4794232767114045e-05,
"loss": 0.9359,
"step": 131500
},
{
"epoch": 1.52,
"learning_rate": 2.4698393347977596e-05,
"loss": 0.9289,
"step": 132000
},
{
"epoch": 1.52,
"learning_rate": 2.460255392884115e-05,
"loss": 0.9361,
"step": 132500
},
{
"epoch": 1.53,
"learning_rate": 2.45067145097047e-05,
"loss": 0.9616,
"step": 133000
},
{
"epoch": 1.54,
"learning_rate": 2.441087509056825e-05,
"loss": 0.9308,
"step": 133500
},
{
"epoch": 1.54,
"learning_rate": 2.4315035671431805e-05,
"loss": 0.9437,
"step": 134000
},
{
"epoch": 1.55,
"learning_rate": 2.4219196252295356e-05,
"loss": 0.9443,
"step": 134500
},
{
"epoch": 1.55,
"learning_rate": 2.4123356833158906e-05,
"loss": 0.9324,
"step": 135000
},
{
"epoch": 1.56,
"learning_rate": 2.4027517414022457e-05,
"loss": 0.9373,
"step": 135500
},
{
"epoch": 1.56,
"learning_rate": 2.3931677994886008e-05,
"loss": 0.9414,
"step": 136000
},
{
"epoch": 1.57,
"learning_rate": 2.3835838575749562e-05,
"loss": 0.9466,
"step": 136500
},
{
"epoch": 1.58,
"learning_rate": 2.3739999156613113e-05,
"loss": 0.9383,
"step": 137000
},
{
"epoch": 1.58,
"learning_rate": 2.3644159737476663e-05,
"loss": 0.9412,
"step": 137500
},
{
"epoch": 1.59,
"learning_rate": 2.3548320318340217e-05,
"loss": 0.9454,
"step": 138000
},
{
"epoch": 1.59,
"learning_rate": 2.3452480899203765e-05,
"loss": 0.9285,
"step": 138500
},
{
"epoch": 1.6,
"learning_rate": 2.335664148006732e-05,
"loss": 0.9261,
"step": 139000
},
{
"epoch": 1.6,
"learning_rate": 2.326080206093087e-05,
"loss": 0.9404,
"step": 139500
},
{
"epoch": 1.61,
"learning_rate": 2.316496264179442e-05,
"loss": 0.9206,
"step": 140000
},
{
"epoch": 1.62,
"learning_rate": 2.3069123222657974e-05,
"loss": 0.9298,
"step": 140500
},
{
"epoch": 1.62,
"learning_rate": 2.2973283803521525e-05,
"loss": 0.9271,
"step": 141000
},
{
"epoch": 1.63,
"learning_rate": 2.287744438438508e-05,
"loss": 0.9431,
"step": 141500
},
{
"epoch": 1.63,
"learning_rate": 2.278160496524863e-05,
"loss": 0.9238,
"step": 142000
},
{
"epoch": 1.64,
"learning_rate": 2.2685765546112177e-05,
"loss": 0.9367,
"step": 142500
},
{
"epoch": 1.64,
"learning_rate": 2.258992612697573e-05,
"loss": 0.9312,
"step": 143000
},
{
"epoch": 1.65,
"learning_rate": 2.249408670783928e-05,
"loss": 0.9355,
"step": 143500
},
{
"epoch": 1.66,
"learning_rate": 2.2398247288702835e-05,
"loss": 0.9381,
"step": 144000
},
{
"epoch": 1.66,
"learning_rate": 2.2302407869566386e-05,
"loss": 0.934,
"step": 144500
},
{
"epoch": 1.67,
"learning_rate": 2.2206568450429937e-05,
"loss": 0.9078,
"step": 145000
},
{
"epoch": 1.67,
"learning_rate": 2.211072903129349e-05,
"loss": 0.9244,
"step": 145500
},
{
"epoch": 1.68,
"learning_rate": 2.2014889612157038e-05,
"loss": 0.9438,
"step": 146000
},
{
"epoch": 1.68,
"learning_rate": 2.191905019302059e-05,
"loss": 0.9224,
"step": 146500
},
{
"epoch": 1.69,
"learning_rate": 2.1823210773884143e-05,
"loss": 0.9323,
"step": 147000
},
{
"epoch": 1.7,
"learning_rate": 2.1727371354747694e-05,
"loss": 0.9333,
"step": 147500
},
{
"epoch": 1.7,
"learning_rate": 2.1631531935611248e-05,
"loss": 0.9267,
"step": 148000
},
{
"epoch": 1.71,
"learning_rate": 2.1535692516474798e-05,
"loss": 0.9096,
"step": 148500
},
{
"epoch": 1.71,
"learning_rate": 2.143985309733835e-05,
"loss": 0.938,
"step": 149000
},
{
"epoch": 1.72,
"learning_rate": 2.13440136782019e-05,
"loss": 0.9271,
"step": 149500
},
{
"epoch": 1.73,
"learning_rate": 2.124817425906545e-05,
"loss": 0.9221,
"step": 150000
},
{
"epoch": 1.73,
"learning_rate": 2.1152334839929004e-05,
"loss": 0.9264,
"step": 150500
},
{
"epoch": 1.74,
"learning_rate": 2.1056495420792555e-05,
"loss": 0.9166,
"step": 151000
},
{
"epoch": 1.74,
"learning_rate": 2.0960656001656106e-05,
"loss": 0.9392,
"step": 151500
},
{
"epoch": 1.75,
"learning_rate": 2.086481658251966e-05,
"loss": 0.9075,
"step": 152000
},
{
"epoch": 1.75,
"learning_rate": 2.076897716338321e-05,
"loss": 0.9103,
"step": 152500
},
{
"epoch": 1.76,
"learning_rate": 2.067313774424676e-05,
"loss": 0.9298,
"step": 153000
},
{
"epoch": 1.77,
"learning_rate": 2.0577298325110312e-05,
"loss": 0.918,
"step": 153500
},
{
"epoch": 1.77,
"learning_rate": 2.0481458905973862e-05,
"loss": 0.9088,
"step": 154000
},
{
"epoch": 1.78,
"learning_rate": 2.0385619486837416e-05,
"loss": 0.9046,
"step": 154500
},
{
"epoch": 1.78,
"learning_rate": 2.0289780067700967e-05,
"loss": 0.9116,
"step": 155000
},
{
"epoch": 1.79,
"learning_rate": 2.0193940648564518e-05,
"loss": 0.9124,
"step": 155500
},
{
"epoch": 1.79,
"learning_rate": 2.0098101229428072e-05,
"loss": 0.9167,
"step": 156000
},
{
"epoch": 1.8,
"learning_rate": 2.000226181029162e-05,
"loss": 0.9143,
"step": 156500
},
{
"epoch": 1.81,
"learning_rate": 1.9906422391155173e-05,
"loss": 0.9047,
"step": 157000
},
{
"epoch": 1.81,
"learning_rate": 1.9810582972018724e-05,
"loss": 0.9129,
"step": 157500
},
{
"epoch": 1.82,
"learning_rate": 1.9714743552882275e-05,
"loss": 0.9019,
"step": 158000
},
{
"epoch": 1.82,
"learning_rate": 1.961890413374583e-05,
"loss": 0.9104,
"step": 158500
},
{
"epoch": 1.83,
"learning_rate": 1.952306471460938e-05,
"loss": 0.9021,
"step": 159000
},
{
"epoch": 1.83,
"learning_rate": 1.942722529547293e-05,
"loss": 0.9206,
"step": 159500
},
{
"epoch": 1.84,
"learning_rate": 1.933138587633648e-05,
"loss": 0.9211,
"step": 160000
},
{
"epoch": 1.85,
"learning_rate": 1.923554645720003e-05,
"loss": 0.9126,
"step": 160500
},
{
"epoch": 1.85,
"learning_rate": 1.9139707038063585e-05,
"loss": 0.9167,
"step": 161000
},
{
"epoch": 1.86,
"learning_rate": 1.9043867618927136e-05,
"loss": 0.9076,
"step": 161500
},
{
"epoch": 1.86,
"learning_rate": 1.8948028199790687e-05,
"loss": 0.908,
"step": 162000
},
{
"epoch": 1.87,
"learning_rate": 1.885218878065424e-05,
"loss": 0.9222,
"step": 162500
},
{
"epoch": 1.87,
"learning_rate": 1.875634936151779e-05,
"loss": 0.9072,
"step": 163000
},
{
"epoch": 1.88,
"learning_rate": 1.8660509942381342e-05,
"loss": 0.9079,
"step": 163500
},
{
"epoch": 1.89,
"learning_rate": 1.8564670523244893e-05,
"loss": 0.9071,
"step": 164000
},
{
"epoch": 1.89,
"learning_rate": 1.8468831104108443e-05,
"loss": 0.9093,
"step": 164500
},
{
"epoch": 1.9,
"learning_rate": 1.8372991684971997e-05,
"loss": 0.8963,
"step": 165000
},
{
"epoch": 1.9,
"learning_rate": 1.8277152265835548e-05,
"loss": 0.8875,
"step": 165500
},
{
"epoch": 1.91,
"learning_rate": 1.81813128466991e-05,
"loss": 0.8972,
"step": 166000
},
{
"epoch": 1.91,
"learning_rate": 1.8085473427562653e-05,
"loss": 0.9027,
"step": 166500
},
{
"epoch": 1.92,
"learning_rate": 1.7989634008426203e-05,
"loss": 0.8843,
"step": 167000
},
{
"epoch": 1.93,
"learning_rate": 1.7893794589289754e-05,
"loss": 0.9142,
"step": 167500
},
{
"epoch": 1.93,
"learning_rate": 1.7797955170153305e-05,
"loss": 0.9125,
"step": 168000
},
{
"epoch": 1.94,
"learning_rate": 1.7702115751016855e-05,
"loss": 0.8859,
"step": 168500
},
{
"epoch": 1.94,
"learning_rate": 1.760627633188041e-05,
"loss": 0.9077,
"step": 169000
},
{
"epoch": 1.95,
"learning_rate": 1.751043691274396e-05,
"loss": 0.911,
"step": 169500
},
{
"epoch": 1.96,
"learning_rate": 1.741459749360751e-05,
"loss": 0.8953,
"step": 170000
},
{
"epoch": 1.96,
"learning_rate": 1.7318758074471065e-05,
"loss": 0.8891,
"step": 170500
},
{
"epoch": 1.97,
"learning_rate": 1.7222918655334612e-05,
"loss": 0.887,
"step": 171000
},
{
"epoch": 1.97,
"learning_rate": 1.7127079236198166e-05,
"loss": 0.8849,
"step": 171500
},
{
"epoch": 1.98,
"learning_rate": 1.7031239817061717e-05,
"loss": 0.8899,
"step": 172000
},
{
"epoch": 1.98,
"learning_rate": 1.6935400397925268e-05,
"loss": 0.8997,
"step": 172500
},
{
"epoch": 1.99,
"learning_rate": 1.683956097878882e-05,
"loss": 0.8807,
"step": 173000
},
{
"epoch": 2.0,
"learning_rate": 1.6743721559652372e-05,
"loss": 0.9092,
"step": 173500
},
{
"epoch": 2.0,
"learning_rate": 1.6647882140515923e-05,
"loss": 0.8584,
"step": 174000
},
{
"epoch": 2.01,
"learning_rate": 1.6552042721379474e-05,
"loss": 0.7926,
"step": 174500
},
{
"epoch": 2.01,
"learning_rate": 1.6456203302243024e-05,
"loss": 0.8028,
"step": 175000
},
{
"epoch": 2.02,
"learning_rate": 1.636036388310658e-05,
"loss": 0.7806,
"step": 175500
},
{
"epoch": 2.02,
"learning_rate": 1.626452446397013e-05,
"loss": 0.7893,
"step": 176000
},
{
"epoch": 2.03,
"learning_rate": 1.616868504483368e-05,
"loss": 0.7922,
"step": 176500
},
{
"epoch": 2.04,
"learning_rate": 1.6072845625697234e-05,
"loss": 0.782,
"step": 177000
},
{
"epoch": 2.04,
"learning_rate": 1.5977006206560784e-05,
"loss": 0.782,
"step": 177500
},
{
"epoch": 2.05,
"learning_rate": 1.5881166787424335e-05,
"loss": 0.7839,
"step": 178000
},
{
"epoch": 2.05,
"learning_rate": 1.5785327368287886e-05,
"loss": 0.7829,
"step": 178500
},
{
"epoch": 2.06,
"learning_rate": 1.5689487949151436e-05,
"loss": 0.784,
"step": 179000
},
{
"epoch": 2.06,
"learning_rate": 1.559364853001499e-05,
"loss": 0.7905,
"step": 179500
},
{
"epoch": 2.07,
"learning_rate": 1.549780911087854e-05,
"loss": 0.7764,
"step": 180000
},
{
"epoch": 2.08,
"learning_rate": 1.5401969691742095e-05,
"loss": 0.8011,
"step": 180500
},
{
"epoch": 2.08,
"learning_rate": 1.5306130272605646e-05,
"loss": 0.7833,
"step": 181000
},
{
"epoch": 2.09,
"learning_rate": 1.5210290853469197e-05,
"loss": 0.7892,
"step": 181500
},
{
"epoch": 2.09,
"learning_rate": 1.5114451434332747e-05,
"loss": 0.794,
"step": 182000
},
{
"epoch": 2.1,
"learning_rate": 1.5018612015196298e-05,
"loss": 0.7772,
"step": 182500
},
{
"epoch": 2.1,
"learning_rate": 1.492277259605985e-05,
"loss": 0.7736,
"step": 183000
},
{
"epoch": 2.11,
"learning_rate": 1.4826933176923403e-05,
"loss": 0.7847,
"step": 183500
},
{
"epoch": 2.12,
"learning_rate": 1.4731093757786953e-05,
"loss": 0.793,
"step": 184000
},
{
"epoch": 2.12,
"learning_rate": 1.4635254338650506e-05,
"loss": 0.7799,
"step": 184500
},
{
"epoch": 2.13,
"learning_rate": 1.4539414919514058e-05,
"loss": 0.7755,
"step": 185000
},
{
"epoch": 2.13,
"learning_rate": 1.4443575500377607e-05,
"loss": 0.7853,
"step": 185500
},
{
"epoch": 2.14,
"learning_rate": 1.434773608124116e-05,
"loss": 0.7751,
"step": 186000
},
{
"epoch": 2.14,
"learning_rate": 1.425189666210471e-05,
"loss": 0.7965,
"step": 186500
},
{
"epoch": 2.15,
"learning_rate": 1.4156057242968262e-05,
"loss": 0.778,
"step": 187000
},
{
"epoch": 2.16,
"learning_rate": 1.4060217823831815e-05,
"loss": 0.7715,
"step": 187500
},
{
"epoch": 2.16,
"learning_rate": 1.3964378404695367e-05,
"loss": 0.7746,
"step": 188000
},
{
"epoch": 2.17,
"learning_rate": 1.3868538985558918e-05,
"loss": 0.7787,
"step": 188500
},
{
"epoch": 2.17,
"learning_rate": 1.3772699566422467e-05,
"loss": 0.7924,
"step": 189000
},
{
"epoch": 2.18,
"learning_rate": 1.3676860147286019e-05,
"loss": 0.7732,
"step": 189500
},
{
"epoch": 2.19,
"learning_rate": 1.3581020728149571e-05,
"loss": 0.7836,
"step": 190000
},
{
"epoch": 2.19,
"learning_rate": 1.3485181309013122e-05,
"loss": 0.7698,
"step": 190500
},
{
"epoch": 2.2,
"learning_rate": 1.3389341889876675e-05,
"loss": 0.782,
"step": 191000
},
{
"epoch": 2.2,
"learning_rate": 1.3293502470740227e-05,
"loss": 0.7732,
"step": 191500
},
{
"epoch": 2.21,
"learning_rate": 1.319766305160378e-05,
"loss": 0.7871,
"step": 192000
},
{
"epoch": 2.21,
"learning_rate": 1.3101823632467328e-05,
"loss": 0.7833,
"step": 192500
},
{
"epoch": 2.22,
"learning_rate": 1.3005984213330879e-05,
"loss": 0.7968,
"step": 193000
},
{
"epoch": 2.23,
"learning_rate": 1.2910144794194431e-05,
"loss": 0.7849,
"step": 193500
},
{
"epoch": 2.23,
"learning_rate": 1.2814305375057984e-05,
"loss": 0.7978,
"step": 194000
},
{
"epoch": 2.24,
"learning_rate": 1.2718465955921536e-05,
"loss": 0.7993,
"step": 194500
},
{
"epoch": 2.24,
"learning_rate": 1.2622626536785087e-05,
"loss": 0.7859,
"step": 195000
},
{
"epoch": 2.25,
"learning_rate": 1.2526787117648639e-05,
"loss": 0.7734,
"step": 195500
},
{
"epoch": 2.25,
"learning_rate": 1.243094769851219e-05,
"loss": 0.7858,
"step": 196000
},
{
"epoch": 2.26,
"learning_rate": 1.2335108279375742e-05,
"loss": 0.7904,
"step": 196500
},
{
"epoch": 2.27,
"learning_rate": 1.2239268860239291e-05,
"loss": 0.7882,
"step": 197000
},
{
"epoch": 2.27,
"learning_rate": 1.2143429441102843e-05,
"loss": 0.7795,
"step": 197500
},
{
"epoch": 2.28,
"learning_rate": 1.2047590021966396e-05,
"loss": 0.7596,
"step": 198000
},
{
"epoch": 2.28,
"learning_rate": 1.1951750602829948e-05,
"loss": 0.7613,
"step": 198500
},
{
"epoch": 2.29,
"learning_rate": 1.1855911183693499e-05,
"loss": 0.7775,
"step": 199000
},
{
"epoch": 2.29,
"learning_rate": 1.176007176455705e-05,
"loss": 0.7779,
"step": 199500
},
{
"epoch": 2.3,
"learning_rate": 1.1664232345420602e-05,
"loss": 0.7957,
"step": 200000
},
{
"epoch": 2.31,
"learning_rate": 1.1568392926284152e-05,
"loss": 0.7648,
"step": 200500
},
{
"epoch": 2.31,
"learning_rate": 1.1472553507147705e-05,
"loss": 0.7709,
"step": 201000
},
{
"epoch": 2.32,
"learning_rate": 1.1376714088011256e-05,
"loss": 0.7995,
"step": 201500
},
{
"epoch": 2.32,
"learning_rate": 1.1280874668874808e-05,
"loss": 0.7712,
"step": 202000
},
{
"epoch": 2.33,
"learning_rate": 1.1185035249738359e-05,
"loss": 0.7739,
"step": 202500
},
{
"epoch": 2.33,
"learning_rate": 1.1089195830601911e-05,
"loss": 0.778,
"step": 203000
},
{
"epoch": 2.34,
"learning_rate": 1.0993356411465462e-05,
"loss": 0.787,
"step": 203500
},
{
"epoch": 2.35,
"learning_rate": 1.0897516992329014e-05,
"loss": 0.7862,
"step": 204000
},
{
"epoch": 2.35,
"learning_rate": 1.0801677573192565e-05,
"loss": 0.7814,
"step": 204500
},
{
"epoch": 2.36,
"learning_rate": 1.0705838154056117e-05,
"loss": 0.7752,
"step": 205000
},
{
"epoch": 2.36,
"learning_rate": 1.0609998734919668e-05,
"loss": 0.7893,
"step": 205500
},
{
"epoch": 2.37,
"learning_rate": 1.0514159315783218e-05,
"loss": 0.774,
"step": 206000
},
{
"epoch": 2.37,
"learning_rate": 1.041831989664677e-05,
"loss": 0.7731,
"step": 206500
},
{
"epoch": 2.38,
"learning_rate": 1.0322480477510323e-05,
"loss": 0.7874,
"step": 207000
},
{
"epoch": 2.39,
"learning_rate": 1.0226641058373874e-05,
"loss": 0.7818,
"step": 207500
},
{
"epoch": 2.39,
"learning_rate": 1.0130801639237424e-05,
"loss": 0.7946,
"step": 208000
},
{
"epoch": 2.4,
"learning_rate": 1.0034962220100977e-05,
"loss": 0.7647,
"step": 208500
},
{
"epoch": 2.4,
"learning_rate": 9.939122800964529e-06,
"loss": 0.7739,
"step": 209000
},
{
"epoch": 2.41,
"learning_rate": 9.84328338182808e-06,
"loss": 0.7721,
"step": 209500
},
{
"epoch": 2.42,
"learning_rate": 9.74744396269163e-06,
"loss": 0.765,
"step": 210000
},
{
"epoch": 2.42,
"learning_rate": 9.651604543555183e-06,
"loss": 0.7783,
"step": 210500
},
{
"epoch": 2.43,
"learning_rate": 9.555765124418735e-06,
"loss": 0.7697,
"step": 211000
},
{
"epoch": 2.43,
"learning_rate": 9.459925705282286e-06,
"loss": 0.7731,
"step": 211500
},
{
"epoch": 2.44,
"learning_rate": 9.364086286145836e-06,
"loss": 0.7833,
"step": 212000
},
{
"epoch": 2.44,
"learning_rate": 9.268246867009389e-06,
"loss": 0.7816,
"step": 212500
},
{
"epoch": 2.45,
"learning_rate": 9.172407447872941e-06,
"loss": 0.7775,
"step": 213000
},
{
"epoch": 2.46,
"learning_rate": 9.076568028736492e-06,
"loss": 0.77,
"step": 213500
},
{
"epoch": 2.46,
"learning_rate": 8.980728609600044e-06,
"loss": 0.7914,
"step": 214000
},
{
"epoch": 2.47,
"learning_rate": 8.884889190463595e-06,
"loss": 0.7726,
"step": 214500
},
{
"epoch": 2.47,
"learning_rate": 8.789049771327146e-06,
"loss": 0.7743,
"step": 215000
},
{
"epoch": 2.48,
"learning_rate": 8.693210352190698e-06,
"loss": 0.7877,
"step": 215500
},
{
"epoch": 2.48,
"learning_rate": 8.59737093305425e-06,
"loss": 0.7607,
"step": 216000
},
{
"epoch": 2.49,
"learning_rate": 8.501531513917801e-06,
"loss": 0.779,
"step": 216500
},
{
"epoch": 2.5,
"learning_rate": 8.405692094781352e-06,
"loss": 0.7781,
"step": 217000
},
{
"epoch": 2.5,
"learning_rate": 8.309852675644904e-06,
"loss": 0.7649,
"step": 217500
},
{
"epoch": 2.51,
"learning_rate": 8.214013256508456e-06,
"loss": 0.7554,
"step": 218000
},
{
"epoch": 2.51,
"learning_rate": 8.118173837372007e-06,
"loss": 0.7413,
"step": 218500
},
{
"epoch": 2.52,
"learning_rate": 8.022334418235558e-06,
"loss": 0.7772,
"step": 219000
},
{
"epoch": 2.52,
"learning_rate": 7.92649499909911e-06,
"loss": 0.7876,
"step": 219500
},
{
"epoch": 2.53,
"learning_rate": 7.830655579962662e-06,
"loss": 0.7572,
"step": 220000
},
{
"epoch": 2.54,
"learning_rate": 7.734816160826213e-06,
"loss": 0.7536,
"step": 220500
},
{
"epoch": 2.54,
"learning_rate": 7.638976741689764e-06,
"loss": 0.7587,
"step": 221000
},
{
"epoch": 2.55,
"learning_rate": 7.543137322553316e-06,
"loss": 0.7718,
"step": 221500
},
{
"epoch": 2.55,
"learning_rate": 7.447297903416868e-06,
"loss": 0.77,
"step": 222000
},
{
"epoch": 2.56,
"learning_rate": 7.351458484280418e-06,
"loss": 0.7493,
"step": 222500
},
{
"epoch": 2.56,
"learning_rate": 7.255619065143971e-06,
"loss": 0.7522,
"step": 223000
},
{
"epoch": 2.57,
"learning_rate": 7.159779646007522e-06,
"loss": 0.7714,
"step": 223500
},
{
"epoch": 2.58,
"learning_rate": 7.063940226871073e-06,
"loss": 0.7682,
"step": 224000
},
{
"epoch": 2.58,
"learning_rate": 6.968100807734624e-06,
"loss": 0.7702,
"step": 224500
},
{
"epoch": 2.59,
"learning_rate": 6.872261388598177e-06,
"loss": 0.7636,
"step": 225000
},
{
"epoch": 2.59,
"learning_rate": 6.776421969461728e-06,
"loss": 0.767,
"step": 225500
},
{
"epoch": 2.6,
"learning_rate": 6.680582550325279e-06,
"loss": 0.7633,
"step": 226000
},
{
"epoch": 2.6,
"learning_rate": 6.5847431311888304e-06,
"loss": 0.7539,
"step": 226500
},
{
"epoch": 2.61,
"learning_rate": 6.488903712052383e-06,
"loss": 0.7555,
"step": 227000
},
{
"epoch": 2.62,
"learning_rate": 6.3930642929159335e-06,
"loss": 0.7587,
"step": 227500
},
{
"epoch": 2.62,
"learning_rate": 6.297224873779485e-06,
"loss": 0.7676,
"step": 228000
},
{
"epoch": 2.63,
"learning_rate": 6.201385454643037e-06,
"loss": 0.7547,
"step": 228500
},
{
"epoch": 2.63,
"learning_rate": 6.105546035506588e-06,
"loss": 0.7619,
"step": 229000
},
{
"epoch": 2.64,
"learning_rate": 6.00970661637014e-06,
"loss": 0.7683,
"step": 229500
},
{
"epoch": 2.65,
"learning_rate": 5.913867197233691e-06,
"loss": 0.7539,
"step": 230000
},
{
"epoch": 2.65,
"learning_rate": 5.818027778097243e-06,
"loss": 0.7586,
"step": 230500
},
{
"epoch": 2.66,
"learning_rate": 5.722188358960794e-06,
"loss": 0.7735,
"step": 231000
},
{
"epoch": 2.66,
"learning_rate": 5.626348939824346e-06,
"loss": 0.7663,
"step": 231500
},
{
"epoch": 2.67,
"learning_rate": 5.530509520687897e-06,
"loss": 0.7616,
"step": 232000
},
{
"epoch": 2.67,
"learning_rate": 5.434670101551449e-06,
"loss": 0.7814,
"step": 232500
},
{
"epoch": 2.68,
"learning_rate": 5.338830682415e-06,
"loss": 0.7697,
"step": 233000
},
{
"epoch": 2.69,
"learning_rate": 5.242991263278552e-06,
"loss": 0.7387,
"step": 233500
},
{
"epoch": 2.69,
"learning_rate": 5.147151844142103e-06,
"loss": 0.7636,
"step": 234000
},
{
"epoch": 2.7,
"learning_rate": 5.051312425005655e-06,
"loss": 0.7659,
"step": 234500
},
{
"epoch": 2.7,
"learning_rate": 4.955473005869206e-06,
"loss": 0.7564,
"step": 235000
},
{
"epoch": 2.71,
"learning_rate": 4.859633586732758e-06,
"loss": 0.7589,
"step": 235500
},
{
"epoch": 2.71,
"learning_rate": 4.763794167596309e-06,
"loss": 0.7683,
"step": 236000
},
{
"epoch": 2.72,
"learning_rate": 4.667954748459861e-06,
"loss": 0.7556,
"step": 236500
},
{
"epoch": 2.73,
"learning_rate": 4.572115329323412e-06,
"loss": 0.7479,
"step": 237000
},
{
"epoch": 2.73,
"learning_rate": 4.476275910186964e-06,
"loss": 0.7548,
"step": 237500
},
{
"epoch": 2.74,
"learning_rate": 4.380436491050515e-06,
"loss": 0.7592,
"step": 238000
},
{
"epoch": 2.74,
"learning_rate": 4.284597071914067e-06,
"loss": 0.7542,
"step": 238500
},
{
"epoch": 2.75,
"learning_rate": 4.188757652777618e-06,
"loss": 0.773,
"step": 239000
},
{
"epoch": 2.75,
"learning_rate": 4.09291823364117e-06,
"loss": 0.7506,
"step": 239500
},
{
"epoch": 2.76,
"learning_rate": 3.997078814504721e-06,
"loss": 0.7489,
"step": 240000
},
{
"epoch": 2.77,
"learning_rate": 3.901239395368272e-06,
"loss": 0.738,
"step": 240500
},
{
"epoch": 2.77,
"learning_rate": 3.8053999762318244e-06,
"loss": 0.7563,
"step": 241000
},
{
"epoch": 2.78,
"learning_rate": 3.7095605570953755e-06,
"loss": 0.7671,
"step": 241500
},
{
"epoch": 2.78,
"learning_rate": 3.6137211379589274e-06,
"loss": 0.7705,
"step": 242000
},
{
"epoch": 2.79,
"learning_rate": 3.5178817188224785e-06,
"loss": 0.7552,
"step": 242500
},
{
"epoch": 2.79,
"learning_rate": 3.4220422996860304e-06,
"loss": 0.7537,
"step": 243000
},
{
"epoch": 2.8,
"learning_rate": 3.3262028805495815e-06,
"loss": 0.7624,
"step": 243500
},
{
"epoch": 2.81,
"learning_rate": 3.2303634614131335e-06,
"loss": 0.7334,
"step": 244000
},
{
"epoch": 2.81,
"learning_rate": 3.1345240422766846e-06,
"loss": 0.7669,
"step": 244500
},
{
"epoch": 2.82,
"learning_rate": 3.038684623140236e-06,
"loss": 0.7512,
"step": 245000
},
{
"epoch": 2.82,
"learning_rate": 2.9428452040037876e-06,
"loss": 0.7587,
"step": 245500
},
{
"epoch": 2.83,
"learning_rate": 2.8470057848673395e-06,
"loss": 0.7635,
"step": 246000
},
{
"epoch": 2.83,
"learning_rate": 2.751166365730891e-06,
"loss": 0.746,
"step": 246500
},
{
"epoch": 2.84,
"learning_rate": 2.655326946594442e-06,
"loss": 0.7512,
"step": 247000
},
{
"epoch": 2.85,
"learning_rate": 2.5594875274579937e-06,
"loss": 0.758,
"step": 247500
},
{
"epoch": 2.85,
"learning_rate": 2.463648108321545e-06,
"loss": 0.7662,
"step": 248000
},
{
"epoch": 2.86,
"learning_rate": 2.3678086891850967e-06,
"loss": 0.7571,
"step": 248500
},
{
"epoch": 2.86,
"learning_rate": 2.271969270048648e-06,
"loss": 0.7453,
"step": 249000
},
{
"epoch": 2.87,
"learning_rate": 2.1761298509121997e-06,
"loss": 0.7494,
"step": 249500
},
{
"epoch": 2.88,
"learning_rate": 2.0802904317757512e-06,
"loss": 0.7574,
"step": 250000
},
{
"epoch": 2.88,
"learning_rate": 1.9844510126393028e-06,
"loss": 0.763,
"step": 250500
},
{
"epoch": 2.89,
"learning_rate": 1.8886115935028543e-06,
"loss": 0.7713,
"step": 251000
},
{
"epoch": 2.89,
"learning_rate": 1.7927721743664056e-06,
"loss": 0.738,
"step": 251500
},
{
"epoch": 2.9,
"learning_rate": 1.696932755229957e-06,
"loss": 0.7483,
"step": 252000
},
{
"epoch": 2.9,
"learning_rate": 1.6010933360935086e-06,
"loss": 0.7458,
"step": 252500
},
{
"epoch": 2.91,
"learning_rate": 1.5052539169570601e-06,
"loss": 0.7664,
"step": 253000
},
{
"epoch": 2.92,
"learning_rate": 1.4094144978206116e-06,
"loss": 0.7516,
"step": 253500
},
{
"epoch": 2.92,
"learning_rate": 1.3135750786841631e-06,
"loss": 0.7628,
"step": 254000
},
{
"epoch": 2.93,
"learning_rate": 1.2177356595477147e-06,
"loss": 0.7455,
"step": 254500
},
{
"epoch": 2.93,
"learning_rate": 1.1218962404112662e-06,
"loss": 0.7601,
"step": 255000
},
{
"epoch": 2.94,
"learning_rate": 1.0260568212748177e-06,
"loss": 0.7471,
"step": 255500
},
{
"epoch": 2.94,
"learning_rate": 9.302174021383692e-07,
"loss": 0.7396,
"step": 256000
},
{
"epoch": 2.95,
"learning_rate": 8.343779830019207e-07,
"loss": 0.7382,
"step": 256500
},
{
"epoch": 2.96,
"learning_rate": 7.385385638654721e-07,
"loss": 0.7501,
"step": 257000
},
{
"epoch": 2.96,
"learning_rate": 6.426991447290237e-07,
"loss": 0.7557,
"step": 257500
},
{
"epoch": 2.97,
"learning_rate": 5.468597255925752e-07,
"loss": 0.7628,
"step": 258000
},
{
"epoch": 2.97,
"learning_rate": 4.5102030645612663e-07,
"loss": 0.752,
"step": 258500
},
{
"epoch": 2.98,
"learning_rate": 3.5518088731967815e-07,
"loss": 0.7496,
"step": 259000
},
{
"epoch": 2.98,
"learning_rate": 2.5934146818322966e-07,
"loss": 0.7588,
"step": 259500
},
{
"epoch": 2.99,
"learning_rate": 1.6350204904678115e-07,
"loss": 0.7423,
"step": 260000
},
{
"epoch": 3.0,
"learning_rate": 6.766262991033264e-08,
"loss": 0.7348,
"step": 260500
},
{
"epoch": 3.0,
"step": 260853,
"total_flos": 3.184114652036137e+17,
"train_loss": 0.9910313752351659,
"train_runtime": 64882.6806,
"train_samples_per_second": 40.204,
"train_steps_per_second": 4.02
}
],
"max_steps": 260853,
"num_train_epochs": 3,
"total_flos": 3.184114652036137e+17,
"trial_name": null,
"trial_params": null
}