AngXLMR-SYN / trainer_state.json
cx-olquinjica's picture
Upload 7 files
d72acb6 verified
raw
history blame
No virus
52.7 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 500,
"global_step": 212490,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.9882347404583746e-05,
"loss": 2.9231,
"step": 500
},
{
"epoch": 0.01,
"learning_rate": 4.9764694809167496e-05,
"loss": 2.1627,
"step": 1000
},
{
"epoch": 0.02,
"learning_rate": 4.964704221375124e-05,
"loss": 1.8699,
"step": 1500
},
{
"epoch": 0.03,
"learning_rate": 4.9529389618334984e-05,
"loss": 1.7267,
"step": 2000
},
{
"epoch": 0.04,
"learning_rate": 4.941173702291873e-05,
"loss": 1.6276,
"step": 2500
},
{
"epoch": 0.04,
"learning_rate": 4.929408442750247e-05,
"loss": 1.5576,
"step": 3000
},
{
"epoch": 0.05,
"learning_rate": 4.9176431832086214e-05,
"loss": 1.5107,
"step": 3500
},
{
"epoch": 0.06,
"learning_rate": 4.905877923666996e-05,
"loss": 1.4644,
"step": 4000
},
{
"epoch": 0.06,
"learning_rate": 4.894112664125371e-05,
"loss": 1.4061,
"step": 4500
},
{
"epoch": 0.07,
"learning_rate": 4.882347404583745e-05,
"loss": 1.3747,
"step": 5000
},
{
"epoch": 0.08,
"learning_rate": 4.8705821450421196e-05,
"loss": 1.3601,
"step": 5500
},
{
"epoch": 0.08,
"learning_rate": 4.8588168855004946e-05,
"loss": 1.3315,
"step": 6000
},
{
"epoch": 0.09,
"learning_rate": 4.847051625958869e-05,
"loss": 1.2922,
"step": 6500
},
{
"epoch": 0.1,
"learning_rate": 4.835286366417243e-05,
"loss": 1.2699,
"step": 7000
},
{
"epoch": 0.11,
"learning_rate": 4.8235211068756184e-05,
"loss": 1.2293,
"step": 7500
},
{
"epoch": 0.11,
"learning_rate": 4.811755847333993e-05,
"loss": 1.227,
"step": 8000
},
{
"epoch": 0.12,
"learning_rate": 4.799990587792367e-05,
"loss": 1.2126,
"step": 8500
},
{
"epoch": 0.13,
"learning_rate": 4.7882253282507415e-05,
"loss": 1.1798,
"step": 9000
},
{
"epoch": 0.13,
"learning_rate": 4.776460068709116e-05,
"loss": 1.1901,
"step": 9500
},
{
"epoch": 0.14,
"learning_rate": 4.76469480916749e-05,
"loss": 1.1668,
"step": 10000
},
{
"epoch": 0.15,
"learning_rate": 4.752929549625865e-05,
"loss": 1.1431,
"step": 10500
},
{
"epoch": 0.16,
"learning_rate": 4.7411642900842396e-05,
"loss": 1.1699,
"step": 11000
},
{
"epoch": 0.16,
"learning_rate": 4.729399030542614e-05,
"loss": 1.1271,
"step": 11500
},
{
"epoch": 0.17,
"learning_rate": 4.717633771000988e-05,
"loss": 1.1251,
"step": 12000
},
{
"epoch": 0.18,
"learning_rate": 4.7058685114593633e-05,
"loss": 1.0963,
"step": 12500
},
{
"epoch": 0.18,
"learning_rate": 4.694103251917738e-05,
"loss": 1.1141,
"step": 13000
},
{
"epoch": 0.19,
"learning_rate": 4.682337992376112e-05,
"loss": 1.0974,
"step": 13500
},
{
"epoch": 0.2,
"learning_rate": 4.6705727328344864e-05,
"loss": 1.0829,
"step": 14000
},
{
"epoch": 0.2,
"learning_rate": 4.658807473292861e-05,
"loss": 1.0893,
"step": 14500
},
{
"epoch": 0.21,
"learning_rate": 4.647042213751235e-05,
"loss": 1.0835,
"step": 15000
},
{
"epoch": 0.22,
"learning_rate": 4.63527695420961e-05,
"loss": 1.0583,
"step": 15500
},
{
"epoch": 0.23,
"learning_rate": 4.6235116946679846e-05,
"loss": 1.0708,
"step": 16000
},
{
"epoch": 0.23,
"learning_rate": 4.611746435126359e-05,
"loss": 1.0695,
"step": 16500
},
{
"epoch": 0.24,
"learning_rate": 4.599981175584734e-05,
"loss": 1.0373,
"step": 17000
},
{
"epoch": 0.25,
"learning_rate": 4.588215916043108e-05,
"loss": 1.02,
"step": 17500
},
{
"epoch": 0.25,
"learning_rate": 4.576450656501483e-05,
"loss": 1.0366,
"step": 18000
},
{
"epoch": 0.26,
"learning_rate": 4.564685396959858e-05,
"loss": 1.0232,
"step": 18500
},
{
"epoch": 0.27,
"learning_rate": 4.552920137418232e-05,
"loss": 0.9946,
"step": 19000
},
{
"epoch": 0.28,
"learning_rate": 4.541154877876606e-05,
"loss": 0.9895,
"step": 19500
},
{
"epoch": 0.28,
"learning_rate": 4.52938961833498e-05,
"loss": 1.0122,
"step": 20000
},
{
"epoch": 0.29,
"learning_rate": 4.517624358793355e-05,
"loss": 1.0048,
"step": 20500
},
{
"epoch": 0.3,
"learning_rate": 4.5058590992517295e-05,
"loss": 0.9837,
"step": 21000
},
{
"epoch": 0.3,
"learning_rate": 4.494093839710104e-05,
"loss": 0.9891,
"step": 21500
},
{
"epoch": 0.31,
"learning_rate": 4.482328580168479e-05,
"loss": 0.9838,
"step": 22000
},
{
"epoch": 0.32,
"learning_rate": 4.470563320626853e-05,
"loss": 0.9881,
"step": 22500
},
{
"epoch": 0.32,
"learning_rate": 4.4587980610852277e-05,
"loss": 0.9387,
"step": 23000
},
{
"epoch": 0.33,
"learning_rate": 4.447032801543603e-05,
"loss": 0.9741,
"step": 23500
},
{
"epoch": 0.34,
"learning_rate": 4.435267542001977e-05,
"loss": 0.9668,
"step": 24000
},
{
"epoch": 0.35,
"learning_rate": 4.4235022824603514e-05,
"loss": 0.9816,
"step": 24500
},
{
"epoch": 0.35,
"learning_rate": 4.411737022918726e-05,
"loss": 0.9479,
"step": 25000
},
{
"epoch": 0.36,
"learning_rate": 4.3999717633771e-05,
"loss": 0.9568,
"step": 25500
},
{
"epoch": 0.37,
"learning_rate": 4.3882065038354745e-05,
"loss": 0.9476,
"step": 26000
},
{
"epoch": 0.37,
"learning_rate": 4.3764412442938495e-05,
"loss": 0.9533,
"step": 26500
},
{
"epoch": 0.38,
"learning_rate": 4.364675984752224e-05,
"loss": 0.9282,
"step": 27000
},
{
"epoch": 0.39,
"learning_rate": 4.352910725210598e-05,
"loss": 0.9454,
"step": 27500
},
{
"epoch": 0.4,
"learning_rate": 4.3411454656689726e-05,
"loss": 0.9418,
"step": 28000
},
{
"epoch": 0.4,
"learning_rate": 4.329380206127348e-05,
"loss": 0.9214,
"step": 28500
},
{
"epoch": 0.41,
"learning_rate": 4.317614946585722e-05,
"loss": 0.9193,
"step": 29000
},
{
"epoch": 0.42,
"learning_rate": 4.3058496870440964e-05,
"loss": 0.8905,
"step": 29500
},
{
"epoch": 0.42,
"learning_rate": 4.294084427502471e-05,
"loss": 0.9192,
"step": 30000
},
{
"epoch": 0.43,
"learning_rate": 4.282319167960845e-05,
"loss": 0.9106,
"step": 30500
},
{
"epoch": 0.44,
"learning_rate": 4.2705539084192195e-05,
"loss": 0.8986,
"step": 31000
},
{
"epoch": 0.44,
"learning_rate": 4.2587886488775945e-05,
"loss": 0.9049,
"step": 31500
},
{
"epoch": 0.45,
"learning_rate": 4.247023389335969e-05,
"loss": 0.8943,
"step": 32000
},
{
"epoch": 0.46,
"learning_rate": 4.235258129794343e-05,
"loss": 0.8912,
"step": 32500
},
{
"epoch": 0.47,
"learning_rate": 4.223492870252718e-05,
"loss": 0.8812,
"step": 33000
},
{
"epoch": 0.47,
"learning_rate": 4.2117276107110926e-05,
"loss": 0.8942,
"step": 33500
},
{
"epoch": 0.48,
"learning_rate": 4.199962351169467e-05,
"loss": 0.8956,
"step": 34000
},
{
"epoch": 0.49,
"learning_rate": 4.1881970916278414e-05,
"loss": 0.8793,
"step": 34500
},
{
"epoch": 0.49,
"learning_rate": 4.1764318320862164e-05,
"loss": 0.8793,
"step": 35000
},
{
"epoch": 0.5,
"learning_rate": 4.164666572544591e-05,
"loss": 0.8847,
"step": 35500
},
{
"epoch": 0.51,
"learning_rate": 4.152901313002965e-05,
"loss": 0.886,
"step": 36000
},
{
"epoch": 0.52,
"learning_rate": 4.1411360534613395e-05,
"loss": 0.8677,
"step": 36500
},
{
"epoch": 0.52,
"learning_rate": 4.129370793919714e-05,
"loss": 0.8751,
"step": 37000
},
{
"epoch": 0.53,
"learning_rate": 4.117605534378088e-05,
"loss": 0.8672,
"step": 37500
},
{
"epoch": 0.54,
"learning_rate": 4.105840274836463e-05,
"loss": 0.8774,
"step": 38000
},
{
"epoch": 0.54,
"learning_rate": 4.0940750152948376e-05,
"loss": 0.8735,
"step": 38500
},
{
"epoch": 0.55,
"learning_rate": 4.082309755753212e-05,
"loss": 0.8639,
"step": 39000
},
{
"epoch": 0.56,
"learning_rate": 4.070544496211587e-05,
"loss": 0.8609,
"step": 39500
},
{
"epoch": 0.56,
"learning_rate": 4.0587792366699614e-05,
"loss": 0.8669,
"step": 40000
},
{
"epoch": 0.57,
"learning_rate": 4.047013977128336e-05,
"loss": 0.8662,
"step": 40500
},
{
"epoch": 0.58,
"learning_rate": 4.03524871758671e-05,
"loss": 0.8522,
"step": 41000
},
{
"epoch": 0.59,
"learning_rate": 4.0234834580450845e-05,
"loss": 0.8609,
"step": 41500
},
{
"epoch": 0.59,
"learning_rate": 4.011718198503459e-05,
"loss": 0.8459,
"step": 42000
},
{
"epoch": 0.6,
"learning_rate": 3.999952938961833e-05,
"loss": 0.8384,
"step": 42500
},
{
"epoch": 0.61,
"learning_rate": 3.988187679420208e-05,
"loss": 0.8335,
"step": 43000
},
{
"epoch": 0.61,
"learning_rate": 3.9764224198785826e-05,
"loss": 0.8567,
"step": 43500
},
{
"epoch": 0.62,
"learning_rate": 3.964657160336957e-05,
"loss": 0.8567,
"step": 44000
},
{
"epoch": 0.63,
"learning_rate": 3.952891900795332e-05,
"loss": 0.8256,
"step": 44500
},
{
"epoch": 0.64,
"learning_rate": 3.9411266412537063e-05,
"loss": 0.8289,
"step": 45000
},
{
"epoch": 0.64,
"learning_rate": 3.929361381712081e-05,
"loss": 0.8406,
"step": 45500
},
{
"epoch": 0.65,
"learning_rate": 3.917596122170456e-05,
"loss": 0.8277,
"step": 46000
},
{
"epoch": 0.66,
"learning_rate": 3.90583086262883e-05,
"loss": 0.807,
"step": 46500
},
{
"epoch": 0.66,
"learning_rate": 3.8940656030872045e-05,
"loss": 0.8062,
"step": 47000
},
{
"epoch": 0.67,
"learning_rate": 3.882300343545579e-05,
"loss": 0.825,
"step": 47500
},
{
"epoch": 0.68,
"learning_rate": 3.870535084003953e-05,
"loss": 0.8083,
"step": 48000
},
{
"epoch": 0.68,
"learning_rate": 3.8587698244623275e-05,
"loss": 0.8222,
"step": 48500
},
{
"epoch": 0.69,
"learning_rate": 3.8470045649207026e-05,
"loss": 0.8262,
"step": 49000
},
{
"epoch": 0.7,
"learning_rate": 3.835239305379077e-05,
"loss": 0.8231,
"step": 49500
},
{
"epoch": 0.71,
"learning_rate": 3.823474045837451e-05,
"loss": 0.8154,
"step": 50000
},
{
"epoch": 0.71,
"learning_rate": 3.811708786295826e-05,
"loss": 0.8021,
"step": 50500
},
{
"epoch": 0.72,
"learning_rate": 3.799943526754201e-05,
"loss": 0.8163,
"step": 51000
},
{
"epoch": 0.73,
"learning_rate": 3.788178267212575e-05,
"loss": 0.8021,
"step": 51500
},
{
"epoch": 0.73,
"learning_rate": 3.7764130076709494e-05,
"loss": 0.7975,
"step": 52000
},
{
"epoch": 0.74,
"learning_rate": 3.764647748129324e-05,
"loss": 0.7899,
"step": 52500
},
{
"epoch": 0.75,
"learning_rate": 3.752882488587698e-05,
"loss": 0.81,
"step": 53000
},
{
"epoch": 0.76,
"learning_rate": 3.7411172290460725e-05,
"loss": 0.8023,
"step": 53500
},
{
"epoch": 0.76,
"learning_rate": 3.7293519695044476e-05,
"loss": 0.7946,
"step": 54000
},
{
"epoch": 0.77,
"learning_rate": 3.717586709962822e-05,
"loss": 0.8181,
"step": 54500
},
{
"epoch": 0.78,
"learning_rate": 3.705821450421196e-05,
"loss": 0.8046,
"step": 55000
},
{
"epoch": 0.78,
"learning_rate": 3.694056190879571e-05,
"loss": 0.809,
"step": 55500
},
{
"epoch": 0.79,
"learning_rate": 3.682290931337946e-05,
"loss": 0.8072,
"step": 56000
},
{
"epoch": 0.8,
"learning_rate": 3.67052567179632e-05,
"loss": 0.7868,
"step": 56500
},
{
"epoch": 0.8,
"learning_rate": 3.6587604122546944e-05,
"loss": 0.7887,
"step": 57000
},
{
"epoch": 0.81,
"learning_rate": 3.6469951527130694e-05,
"loss": 0.7829,
"step": 57500
},
{
"epoch": 0.82,
"learning_rate": 3.635229893171443e-05,
"loss": 0.7685,
"step": 58000
},
{
"epoch": 0.83,
"learning_rate": 3.6234646336298175e-05,
"loss": 0.7888,
"step": 58500
},
{
"epoch": 0.83,
"learning_rate": 3.6116993740881925e-05,
"loss": 0.791,
"step": 59000
},
{
"epoch": 0.84,
"learning_rate": 3.599934114546567e-05,
"loss": 0.7961,
"step": 59500
},
{
"epoch": 0.85,
"learning_rate": 3.588168855004941e-05,
"loss": 0.7733,
"step": 60000
},
{
"epoch": 0.85,
"learning_rate": 3.576403595463316e-05,
"loss": 0.7765,
"step": 60500
},
{
"epoch": 0.86,
"learning_rate": 3.5646383359216907e-05,
"loss": 0.7793,
"step": 61000
},
{
"epoch": 0.87,
"learning_rate": 3.552873076380065e-05,
"loss": 0.7985,
"step": 61500
},
{
"epoch": 0.88,
"learning_rate": 3.54110781683844e-05,
"loss": 0.7671,
"step": 62000
},
{
"epoch": 0.88,
"learning_rate": 3.5293425572968144e-05,
"loss": 0.779,
"step": 62500
},
{
"epoch": 0.89,
"learning_rate": 3.517577297755189e-05,
"loss": 0.7888,
"step": 63000
},
{
"epoch": 0.9,
"learning_rate": 3.505812038213563e-05,
"loss": 0.7731,
"step": 63500
},
{
"epoch": 0.9,
"learning_rate": 3.4940467786719375e-05,
"loss": 0.783,
"step": 64000
},
{
"epoch": 0.91,
"learning_rate": 3.482281519130312e-05,
"loss": 0.7639,
"step": 64500
},
{
"epoch": 0.92,
"learning_rate": 3.470516259588686e-05,
"loss": 0.7659,
"step": 65000
},
{
"epoch": 0.92,
"learning_rate": 3.458751000047061e-05,
"loss": 0.7667,
"step": 65500
},
{
"epoch": 0.93,
"learning_rate": 3.4469857405054356e-05,
"loss": 0.7564,
"step": 66000
},
{
"epoch": 0.94,
"learning_rate": 3.43522048096381e-05,
"loss": 0.7697,
"step": 66500
},
{
"epoch": 0.95,
"learning_rate": 3.423455221422185e-05,
"loss": 0.765,
"step": 67000
},
{
"epoch": 0.95,
"learning_rate": 3.4116899618805594e-05,
"loss": 0.7538,
"step": 67500
},
{
"epoch": 0.96,
"learning_rate": 3.399924702338934e-05,
"loss": 0.7685,
"step": 68000
},
{
"epoch": 0.97,
"learning_rate": 3.388159442797309e-05,
"loss": 0.7635,
"step": 68500
},
{
"epoch": 0.97,
"learning_rate": 3.3763941832556825e-05,
"loss": 0.7547,
"step": 69000
},
{
"epoch": 0.98,
"learning_rate": 3.364628923714057e-05,
"loss": 0.7571,
"step": 69500
},
{
"epoch": 0.99,
"learning_rate": 3.352863664172432e-05,
"loss": 0.7492,
"step": 70000
},
{
"epoch": 1.0,
"learning_rate": 3.341098404630806e-05,
"loss": 0.7469,
"step": 70500
},
{
"epoch": 1.0,
"learning_rate": 3.3293331450891806e-05,
"loss": 0.7685,
"step": 71000
},
{
"epoch": 1.01,
"learning_rate": 3.3175678855475556e-05,
"loss": 0.7542,
"step": 71500
},
{
"epoch": 1.02,
"learning_rate": 3.30580262600593e-05,
"loss": 0.7503,
"step": 72000
},
{
"epoch": 1.02,
"learning_rate": 3.2940373664643044e-05,
"loss": 0.7487,
"step": 72500
},
{
"epoch": 1.03,
"learning_rate": 3.282272106922679e-05,
"loss": 0.7496,
"step": 73000
},
{
"epoch": 1.04,
"learning_rate": 3.270506847381054e-05,
"loss": 0.7461,
"step": 73500
},
{
"epoch": 1.04,
"learning_rate": 3.258741587839428e-05,
"loss": 0.7283,
"step": 74000
},
{
"epoch": 1.05,
"learning_rate": 3.2469763282978025e-05,
"loss": 0.7378,
"step": 74500
},
{
"epoch": 1.06,
"learning_rate": 3.235211068756177e-05,
"loss": 0.731,
"step": 75000
},
{
"epoch": 1.07,
"learning_rate": 3.223445809214551e-05,
"loss": 0.7312,
"step": 75500
},
{
"epoch": 1.07,
"learning_rate": 3.2116805496729256e-05,
"loss": 0.7316,
"step": 76000
},
{
"epoch": 1.08,
"learning_rate": 3.1999152901313006e-05,
"loss": 0.7235,
"step": 76500
},
{
"epoch": 1.09,
"learning_rate": 3.188150030589675e-05,
"loss": 0.736,
"step": 77000
},
{
"epoch": 1.09,
"learning_rate": 3.176384771048049e-05,
"loss": 0.7389,
"step": 77500
},
{
"epoch": 1.1,
"learning_rate": 3.1646195115064244e-05,
"loss": 0.7232,
"step": 78000
},
{
"epoch": 1.11,
"learning_rate": 3.152854251964799e-05,
"loss": 0.744,
"step": 78500
},
{
"epoch": 1.12,
"learning_rate": 3.141088992423173e-05,
"loss": 0.734,
"step": 79000
},
{
"epoch": 1.12,
"learning_rate": 3.1293237328815475e-05,
"loss": 0.7253,
"step": 79500
},
{
"epoch": 1.13,
"learning_rate": 3.117558473339922e-05,
"loss": 0.7222,
"step": 80000
},
{
"epoch": 1.14,
"learning_rate": 3.105793213798296e-05,
"loss": 0.7347,
"step": 80500
},
{
"epoch": 1.14,
"learning_rate": 3.0940279542566705e-05,
"loss": 0.7022,
"step": 81000
},
{
"epoch": 1.15,
"learning_rate": 3.0822626947150456e-05,
"loss": 0.7324,
"step": 81500
},
{
"epoch": 1.16,
"learning_rate": 3.07049743517342e-05,
"loss": 0.7156,
"step": 82000
},
{
"epoch": 1.16,
"learning_rate": 3.058732175631794e-05,
"loss": 0.7259,
"step": 82500
},
{
"epoch": 1.17,
"learning_rate": 3.0469669160901693e-05,
"loss": 0.7161,
"step": 83000
},
{
"epoch": 1.18,
"learning_rate": 3.0352016565485437e-05,
"loss": 0.7265,
"step": 83500
},
{
"epoch": 1.19,
"learning_rate": 3.023436397006918e-05,
"loss": 0.729,
"step": 84000
},
{
"epoch": 1.19,
"learning_rate": 3.0116711374652928e-05,
"loss": 0.7185,
"step": 84500
},
{
"epoch": 1.2,
"learning_rate": 2.999905877923667e-05,
"loss": 0.7164,
"step": 85000
},
{
"epoch": 1.21,
"learning_rate": 2.9881406183820415e-05,
"loss": 0.6989,
"step": 85500
},
{
"epoch": 1.21,
"learning_rate": 2.9763753588404165e-05,
"loss": 0.7204,
"step": 86000
},
{
"epoch": 1.22,
"learning_rate": 2.964610099298791e-05,
"loss": 0.7228,
"step": 86500
},
{
"epoch": 1.23,
"learning_rate": 2.952844839757165e-05,
"loss": 0.7076,
"step": 87000
},
{
"epoch": 1.24,
"learning_rate": 2.9410795802155393e-05,
"loss": 0.7198,
"step": 87500
},
{
"epoch": 1.24,
"learning_rate": 2.9293143206739143e-05,
"loss": 0.7035,
"step": 88000
},
{
"epoch": 1.25,
"learning_rate": 2.9175490611322887e-05,
"loss": 0.707,
"step": 88500
},
{
"epoch": 1.26,
"learning_rate": 2.905783801590663e-05,
"loss": 0.7194,
"step": 89000
},
{
"epoch": 1.26,
"learning_rate": 2.8940185420490377e-05,
"loss": 0.6997,
"step": 89500
},
{
"epoch": 1.27,
"learning_rate": 2.882253282507412e-05,
"loss": 0.714,
"step": 90000
},
{
"epoch": 1.28,
"learning_rate": 2.8704880229657865e-05,
"loss": 0.6986,
"step": 90500
},
{
"epoch": 1.28,
"learning_rate": 2.8587227634241615e-05,
"loss": 0.6982,
"step": 91000
},
{
"epoch": 1.29,
"learning_rate": 2.846957503882536e-05,
"loss": 0.6838,
"step": 91500
},
{
"epoch": 1.3,
"learning_rate": 2.8351922443409102e-05,
"loss": 0.711,
"step": 92000
},
{
"epoch": 1.31,
"learning_rate": 2.823426984799285e-05,
"loss": 0.7111,
"step": 92500
},
{
"epoch": 1.31,
"learning_rate": 2.8116617252576593e-05,
"loss": 0.7183,
"step": 93000
},
{
"epoch": 1.32,
"learning_rate": 2.7998964657160337e-05,
"loss": 0.7133,
"step": 93500
},
{
"epoch": 1.33,
"learning_rate": 2.7881312061744087e-05,
"loss": 0.717,
"step": 94000
},
{
"epoch": 1.33,
"learning_rate": 2.776365946632783e-05,
"loss": 0.7153,
"step": 94500
},
{
"epoch": 1.34,
"learning_rate": 2.7646006870911574e-05,
"loss": 0.7085,
"step": 95000
},
{
"epoch": 1.35,
"learning_rate": 2.7528354275495318e-05,
"loss": 0.7139,
"step": 95500
},
{
"epoch": 1.36,
"learning_rate": 2.7410701680079065e-05,
"loss": 0.6945,
"step": 96000
},
{
"epoch": 1.36,
"learning_rate": 2.729304908466281e-05,
"loss": 0.7022,
"step": 96500
},
{
"epoch": 1.37,
"learning_rate": 2.7175396489246552e-05,
"loss": 0.689,
"step": 97000
},
{
"epoch": 1.38,
"learning_rate": 2.7057743893830302e-05,
"loss": 0.6878,
"step": 97500
},
{
"epoch": 1.38,
"learning_rate": 2.6940091298414043e-05,
"loss": 0.6872,
"step": 98000
},
{
"epoch": 1.39,
"learning_rate": 2.6822438702997786e-05,
"loss": 0.6927,
"step": 98500
},
{
"epoch": 1.4,
"learning_rate": 2.6704786107581537e-05,
"loss": 0.6944,
"step": 99000
},
{
"epoch": 1.4,
"learning_rate": 2.658713351216528e-05,
"loss": 0.6956,
"step": 99500
},
{
"epoch": 1.41,
"learning_rate": 2.6469480916749024e-05,
"loss": 0.6888,
"step": 100000
},
{
"epoch": 1.42,
"learning_rate": 2.635182832133277e-05,
"loss": 0.6881,
"step": 100500
},
{
"epoch": 1.43,
"learning_rate": 2.6234175725916514e-05,
"loss": 0.6932,
"step": 101000
},
{
"epoch": 1.43,
"learning_rate": 2.6116523130500258e-05,
"loss": 0.691,
"step": 101500
},
{
"epoch": 1.44,
"learning_rate": 2.599887053508401e-05,
"loss": 0.6786,
"step": 102000
},
{
"epoch": 1.45,
"learning_rate": 2.5881217939667752e-05,
"loss": 0.6939,
"step": 102500
},
{
"epoch": 1.45,
"learning_rate": 2.5763565344251496e-05,
"loss": 0.6899,
"step": 103000
},
{
"epoch": 1.46,
"learning_rate": 2.564591274883524e-05,
"loss": 0.694,
"step": 103500
},
{
"epoch": 1.47,
"learning_rate": 2.5528260153418986e-05,
"loss": 0.6734,
"step": 104000
},
{
"epoch": 1.48,
"learning_rate": 2.541060755800273e-05,
"loss": 0.6901,
"step": 104500
},
{
"epoch": 1.48,
"learning_rate": 2.5292954962586474e-05,
"loss": 0.6726,
"step": 105000
},
{
"epoch": 1.49,
"learning_rate": 2.5175302367170224e-05,
"loss": 0.6855,
"step": 105500
},
{
"epoch": 1.5,
"learning_rate": 2.5057649771753968e-05,
"loss": 0.6883,
"step": 106000
},
{
"epoch": 1.5,
"learning_rate": 2.493999717633771e-05,
"loss": 0.6785,
"step": 106500
},
{
"epoch": 1.51,
"learning_rate": 2.4822344580921455e-05,
"loss": 0.6753,
"step": 107000
},
{
"epoch": 1.52,
"learning_rate": 2.4704691985505202e-05,
"loss": 0.6728,
"step": 107500
},
{
"epoch": 1.52,
"learning_rate": 2.458703939008895e-05,
"loss": 0.6804,
"step": 108000
},
{
"epoch": 1.53,
"learning_rate": 2.4469386794672692e-05,
"loss": 0.6792,
"step": 108500
},
{
"epoch": 1.54,
"learning_rate": 2.4351734199256436e-05,
"loss": 0.6722,
"step": 109000
},
{
"epoch": 1.55,
"learning_rate": 2.423408160384018e-05,
"loss": 0.6629,
"step": 109500
},
{
"epoch": 1.55,
"learning_rate": 2.4116429008423927e-05,
"loss": 0.664,
"step": 110000
},
{
"epoch": 1.56,
"learning_rate": 2.3998776413007674e-05,
"loss": 0.6612,
"step": 110500
},
{
"epoch": 1.57,
"learning_rate": 2.3881123817591417e-05,
"loss": 0.6662,
"step": 111000
},
{
"epoch": 1.57,
"learning_rate": 2.3763471222175164e-05,
"loss": 0.6791,
"step": 111500
},
{
"epoch": 1.58,
"learning_rate": 2.3645818626758908e-05,
"loss": 0.6658,
"step": 112000
},
{
"epoch": 1.59,
"learning_rate": 2.352816603134265e-05,
"loss": 0.6676,
"step": 112500
},
{
"epoch": 1.6,
"learning_rate": 2.34105134359264e-05,
"loss": 0.6798,
"step": 113000
},
{
"epoch": 1.6,
"learning_rate": 2.3292860840510142e-05,
"loss": 0.6786,
"step": 113500
},
{
"epoch": 1.61,
"learning_rate": 2.317520824509389e-05,
"loss": 0.661,
"step": 114000
},
{
"epoch": 1.62,
"learning_rate": 2.3057555649677633e-05,
"loss": 0.6662,
"step": 114500
},
{
"epoch": 1.62,
"learning_rate": 2.2939903054261376e-05,
"loss": 0.6649,
"step": 115000
},
{
"epoch": 1.63,
"learning_rate": 2.2822250458845123e-05,
"loss": 0.657,
"step": 115500
},
{
"epoch": 1.64,
"learning_rate": 2.2704597863428867e-05,
"loss": 0.6664,
"step": 116000
},
{
"epoch": 1.64,
"learning_rate": 2.2586945268012614e-05,
"loss": 0.6765,
"step": 116500
},
{
"epoch": 1.65,
"learning_rate": 2.2469292672596358e-05,
"loss": 0.664,
"step": 117000
},
{
"epoch": 1.66,
"learning_rate": 2.23516400771801e-05,
"loss": 0.6664,
"step": 117500
},
{
"epoch": 1.67,
"learning_rate": 2.2233987481763848e-05,
"loss": 0.6532,
"step": 118000
},
{
"epoch": 1.67,
"learning_rate": 2.2116334886347595e-05,
"loss": 0.6687,
"step": 118500
},
{
"epoch": 1.68,
"learning_rate": 2.199868229093134e-05,
"loss": 0.6621,
"step": 119000
},
{
"epoch": 1.69,
"learning_rate": 2.1881029695515086e-05,
"loss": 0.6671,
"step": 119500
},
{
"epoch": 1.69,
"learning_rate": 2.176337710009883e-05,
"loss": 0.6619,
"step": 120000
},
{
"epoch": 1.7,
"learning_rate": 2.1645724504682573e-05,
"loss": 0.6549,
"step": 120500
},
{
"epoch": 1.71,
"learning_rate": 2.152807190926632e-05,
"loss": 0.6668,
"step": 121000
},
{
"epoch": 1.72,
"learning_rate": 2.1410419313850064e-05,
"loss": 0.6439,
"step": 121500
},
{
"epoch": 1.72,
"learning_rate": 2.129276671843381e-05,
"loss": 0.6554,
"step": 122000
},
{
"epoch": 1.73,
"learning_rate": 2.1175114123017554e-05,
"loss": 0.6395,
"step": 122500
},
{
"epoch": 1.74,
"learning_rate": 2.1057461527601298e-05,
"loss": 0.6482,
"step": 123000
},
{
"epoch": 1.74,
"learning_rate": 2.0939808932185045e-05,
"loss": 0.6607,
"step": 123500
},
{
"epoch": 1.75,
"learning_rate": 2.082215633676879e-05,
"loss": 0.6508,
"step": 124000
},
{
"epoch": 1.76,
"learning_rate": 2.0704503741352536e-05,
"loss": 0.6438,
"step": 124500
},
{
"epoch": 1.76,
"learning_rate": 2.0586851145936283e-05,
"loss": 0.658,
"step": 125000
},
{
"epoch": 1.77,
"learning_rate": 2.0469198550520026e-05,
"loss": 0.6409,
"step": 125500
},
{
"epoch": 1.78,
"learning_rate": 2.035154595510377e-05,
"loss": 0.6439,
"step": 126000
},
{
"epoch": 1.79,
"learning_rate": 2.0233893359687517e-05,
"loss": 0.6674,
"step": 126500
},
{
"epoch": 1.79,
"learning_rate": 2.011624076427126e-05,
"loss": 0.66,
"step": 127000
},
{
"epoch": 1.8,
"learning_rate": 1.9998588168855007e-05,
"loss": 0.6576,
"step": 127500
},
{
"epoch": 1.81,
"learning_rate": 1.988093557343875e-05,
"loss": 0.6445,
"step": 128000
},
{
"epoch": 1.81,
"learning_rate": 1.9763282978022495e-05,
"loss": 0.6542,
"step": 128500
},
{
"epoch": 1.82,
"learning_rate": 1.964563038260624e-05,
"loss": 0.6412,
"step": 129000
},
{
"epoch": 1.83,
"learning_rate": 1.9527977787189985e-05,
"loss": 0.6496,
"step": 129500
},
{
"epoch": 1.84,
"learning_rate": 1.9410325191773732e-05,
"loss": 0.6423,
"step": 130000
},
{
"epoch": 1.84,
"learning_rate": 1.929267259635748e-05,
"loss": 0.6437,
"step": 130500
},
{
"epoch": 1.85,
"learning_rate": 1.917502000094122e-05,
"loss": 0.6467,
"step": 131000
},
{
"epoch": 1.86,
"learning_rate": 1.9057367405524967e-05,
"loss": 0.6481,
"step": 131500
},
{
"epoch": 1.86,
"learning_rate": 1.893971481010871e-05,
"loss": 0.6442,
"step": 132000
},
{
"epoch": 1.87,
"learning_rate": 1.8822062214692457e-05,
"loss": 0.6345,
"step": 132500
},
{
"epoch": 1.88,
"learning_rate": 1.8704409619276204e-05,
"loss": 0.642,
"step": 133000
},
{
"epoch": 1.88,
"learning_rate": 1.8586757023859948e-05,
"loss": 0.6373,
"step": 133500
},
{
"epoch": 1.89,
"learning_rate": 1.846910442844369e-05,
"loss": 0.6461,
"step": 134000
},
{
"epoch": 1.9,
"learning_rate": 1.835145183302744e-05,
"loss": 0.6429,
"step": 134500
},
{
"epoch": 1.91,
"learning_rate": 1.8233799237611182e-05,
"loss": 0.6389,
"step": 135000
},
{
"epoch": 1.91,
"learning_rate": 1.811614664219493e-05,
"loss": 0.6395,
"step": 135500
},
{
"epoch": 1.92,
"learning_rate": 1.7998494046778673e-05,
"loss": 0.6502,
"step": 136000
},
{
"epoch": 1.93,
"learning_rate": 1.7880841451362416e-05,
"loss": 0.6417,
"step": 136500
},
{
"epoch": 1.93,
"learning_rate": 1.7763188855946163e-05,
"loss": 0.6376,
"step": 137000
},
{
"epoch": 1.94,
"learning_rate": 1.7645536260529907e-05,
"loss": 0.6368,
"step": 137500
},
{
"epoch": 1.95,
"learning_rate": 1.7527883665113654e-05,
"loss": 0.638,
"step": 138000
},
{
"epoch": 1.96,
"learning_rate": 1.74102310696974e-05,
"loss": 0.6508,
"step": 138500
},
{
"epoch": 1.96,
"learning_rate": 1.7292578474281145e-05,
"loss": 0.6311,
"step": 139000
},
{
"epoch": 1.97,
"learning_rate": 1.7174925878864888e-05,
"loss": 0.641,
"step": 139500
},
{
"epoch": 1.98,
"learning_rate": 1.7057273283448632e-05,
"loss": 0.638,
"step": 140000
},
{
"epoch": 1.98,
"learning_rate": 1.693962068803238e-05,
"loss": 0.6176,
"step": 140500
},
{
"epoch": 1.99,
"learning_rate": 1.6821968092616126e-05,
"loss": 0.6346,
"step": 141000
},
{
"epoch": 2.0,
"learning_rate": 1.670431549719987e-05,
"loss": 0.6348,
"step": 141500
},
{
"epoch": 2.0,
"learning_rate": 1.6586662901783613e-05,
"loss": 0.6454,
"step": 142000
},
{
"epoch": 2.01,
"learning_rate": 1.6469010306367357e-05,
"loss": 0.6237,
"step": 142500
},
{
"epoch": 2.02,
"learning_rate": 1.6351357710951104e-05,
"loss": 0.622,
"step": 143000
},
{
"epoch": 2.03,
"learning_rate": 1.623370511553485e-05,
"loss": 0.6112,
"step": 143500
},
{
"epoch": 2.03,
"learning_rate": 1.6116052520118594e-05,
"loss": 0.6288,
"step": 144000
},
{
"epoch": 2.04,
"learning_rate": 1.599839992470234e-05,
"loss": 0.641,
"step": 144500
},
{
"epoch": 2.05,
"learning_rate": 1.5880747329286085e-05,
"loss": 0.6298,
"step": 145000
},
{
"epoch": 2.05,
"learning_rate": 1.576309473386983e-05,
"loss": 0.6144,
"step": 145500
},
{
"epoch": 2.06,
"learning_rate": 1.5645442138453575e-05,
"loss": 0.6261,
"step": 146000
},
{
"epoch": 2.07,
"learning_rate": 1.552778954303732e-05,
"loss": 0.6195,
"step": 146500
},
{
"epoch": 2.08,
"learning_rate": 1.5410136947621066e-05,
"loss": 0.6249,
"step": 147000
},
{
"epoch": 2.08,
"learning_rate": 1.529248435220481e-05,
"loss": 0.6395,
"step": 147500
},
{
"epoch": 2.09,
"learning_rate": 1.5174831756788555e-05,
"loss": 0.6273,
"step": 148000
},
{
"epoch": 2.1,
"learning_rate": 1.50571791613723e-05,
"loss": 0.6279,
"step": 148500
},
{
"epoch": 2.1,
"learning_rate": 1.4939526565956047e-05,
"loss": 0.6157,
"step": 149000
},
{
"epoch": 2.11,
"learning_rate": 1.4821873970539791e-05,
"loss": 0.6238,
"step": 149500
},
{
"epoch": 2.12,
"learning_rate": 1.4704221375123536e-05,
"loss": 0.6222,
"step": 150000
},
{
"epoch": 2.12,
"learning_rate": 1.458656877970728e-05,
"loss": 0.6292,
"step": 150500
},
{
"epoch": 2.13,
"learning_rate": 1.4468916184291025e-05,
"loss": 0.6159,
"step": 151000
},
{
"epoch": 2.14,
"learning_rate": 1.4351263588874772e-05,
"loss": 0.6197,
"step": 151500
},
{
"epoch": 2.15,
"learning_rate": 1.4233610993458516e-05,
"loss": 0.6247,
"step": 152000
},
{
"epoch": 2.15,
"learning_rate": 1.4115958398042261e-05,
"loss": 0.6152,
"step": 152500
},
{
"epoch": 2.16,
"learning_rate": 1.3998305802626008e-05,
"loss": 0.6279,
"step": 153000
},
{
"epoch": 2.17,
"learning_rate": 1.3880653207209752e-05,
"loss": 0.6107,
"step": 153500
},
{
"epoch": 2.17,
"learning_rate": 1.3763000611793497e-05,
"loss": 0.62,
"step": 154000
},
{
"epoch": 2.18,
"learning_rate": 1.364534801637724e-05,
"loss": 0.6009,
"step": 154500
},
{
"epoch": 2.19,
"learning_rate": 1.3527695420960988e-05,
"loss": 0.6217,
"step": 155000
},
{
"epoch": 2.2,
"learning_rate": 1.3410042825544733e-05,
"loss": 0.613,
"step": 155500
},
{
"epoch": 2.2,
"learning_rate": 1.3292390230128477e-05,
"loss": 0.6257,
"step": 156000
},
{
"epoch": 2.21,
"learning_rate": 1.3174737634712222e-05,
"loss": 0.6257,
"step": 156500
},
{
"epoch": 2.22,
"learning_rate": 1.3057085039295969e-05,
"loss": 0.6184,
"step": 157000
},
{
"epoch": 2.22,
"learning_rate": 1.2939432443879713e-05,
"loss": 0.6118,
"step": 157500
},
{
"epoch": 2.23,
"learning_rate": 1.2821779848463458e-05,
"loss": 0.5928,
"step": 158000
},
{
"epoch": 2.24,
"learning_rate": 1.2704127253047201e-05,
"loss": 0.6092,
"step": 158500
},
{
"epoch": 2.24,
"learning_rate": 1.2586474657630948e-05,
"loss": 0.6055,
"step": 159000
},
{
"epoch": 2.25,
"learning_rate": 1.2468822062214692e-05,
"loss": 0.6102,
"step": 159500
},
{
"epoch": 2.26,
"learning_rate": 1.2351169466798439e-05,
"loss": 0.6194,
"step": 160000
},
{
"epoch": 2.27,
"learning_rate": 1.2233516871382183e-05,
"loss": 0.6108,
"step": 160500
},
{
"epoch": 2.27,
"learning_rate": 1.2115864275965928e-05,
"loss": 0.6091,
"step": 161000
},
{
"epoch": 2.28,
"learning_rate": 1.1998211680549673e-05,
"loss": 0.6019,
"step": 161500
},
{
"epoch": 2.29,
"learning_rate": 1.1880559085133419e-05,
"loss": 0.6176,
"step": 162000
},
{
"epoch": 2.29,
"learning_rate": 1.1762906489717164e-05,
"loss": 0.6149,
"step": 162500
},
{
"epoch": 2.3,
"learning_rate": 1.164525389430091e-05,
"loss": 0.5845,
"step": 163000
},
{
"epoch": 2.31,
"learning_rate": 1.1527601298884653e-05,
"loss": 0.6121,
"step": 163500
},
{
"epoch": 2.32,
"learning_rate": 1.14099487034684e-05,
"loss": 0.6092,
"step": 164000
},
{
"epoch": 2.32,
"learning_rate": 1.1292296108052145e-05,
"loss": 0.606,
"step": 164500
},
{
"epoch": 2.33,
"learning_rate": 1.1174643512635889e-05,
"loss": 0.6011,
"step": 165000
},
{
"epoch": 2.34,
"learning_rate": 1.1056990917219634e-05,
"loss": 0.6015,
"step": 165500
},
{
"epoch": 2.34,
"learning_rate": 1.093933832180338e-05,
"loss": 0.6132,
"step": 166000
},
{
"epoch": 2.35,
"learning_rate": 1.0821685726387125e-05,
"loss": 0.6025,
"step": 166500
},
{
"epoch": 2.36,
"learning_rate": 1.070403313097087e-05,
"loss": 0.6099,
"step": 167000
},
{
"epoch": 2.36,
"learning_rate": 1.0586380535554614e-05,
"loss": 0.6172,
"step": 167500
},
{
"epoch": 2.37,
"learning_rate": 1.046872794013836e-05,
"loss": 0.6132,
"step": 168000
},
{
"epoch": 2.38,
"learning_rate": 1.0351075344722106e-05,
"loss": 0.5979,
"step": 168500
},
{
"epoch": 2.39,
"learning_rate": 1.023342274930585e-05,
"loss": 0.6012,
"step": 169000
},
{
"epoch": 2.39,
"learning_rate": 1.0115770153889595e-05,
"loss": 0.5921,
"step": 169500
},
{
"epoch": 2.4,
"learning_rate": 9.998117558473342e-06,
"loss": 0.6039,
"step": 170000
},
{
"epoch": 2.41,
"learning_rate": 9.880464963057086e-06,
"loss": 0.6064,
"step": 170500
},
{
"epoch": 2.41,
"learning_rate": 9.76281236764083e-06,
"loss": 0.6084,
"step": 171000
},
{
"epoch": 2.42,
"learning_rate": 9.645159772224576e-06,
"loss": 0.5933,
"step": 171500
},
{
"epoch": 2.43,
"learning_rate": 9.52750717680832e-06,
"loss": 0.6008,
"step": 172000
},
{
"epoch": 2.44,
"learning_rate": 9.409854581392067e-06,
"loss": 0.6048,
"step": 172500
},
{
"epoch": 2.44,
"learning_rate": 9.29220198597581e-06,
"loss": 0.5933,
"step": 173000
},
{
"epoch": 2.45,
"learning_rate": 9.174549390559556e-06,
"loss": 0.5908,
"step": 173500
},
{
"epoch": 2.46,
"learning_rate": 9.056896795143301e-06,
"loss": 0.5921,
"step": 174000
},
{
"epoch": 2.46,
"learning_rate": 8.939244199727046e-06,
"loss": 0.6012,
"step": 174500
},
{
"epoch": 2.47,
"learning_rate": 8.821591604310792e-06,
"loss": 0.5988,
"step": 175000
},
{
"epoch": 2.48,
"learning_rate": 8.703939008894537e-06,
"loss": 0.6019,
"step": 175500
},
{
"epoch": 2.48,
"learning_rate": 8.58628641347828e-06,
"loss": 0.5915,
"step": 176000
},
{
"epoch": 2.49,
"learning_rate": 8.468633818062028e-06,
"loss": 0.6104,
"step": 176500
},
{
"epoch": 2.5,
"learning_rate": 8.350981222645773e-06,
"loss": 0.6093,
"step": 177000
},
{
"epoch": 2.51,
"learning_rate": 8.233328627229516e-06,
"loss": 0.6084,
"step": 177500
},
{
"epoch": 2.51,
"learning_rate": 8.115676031813262e-06,
"loss": 0.5989,
"step": 178000
},
{
"epoch": 2.52,
"learning_rate": 7.998023436397007e-06,
"loss": 0.6088,
"step": 178500
},
{
"epoch": 2.53,
"learning_rate": 7.880370840980752e-06,
"loss": 0.5889,
"step": 179000
},
{
"epoch": 2.53,
"learning_rate": 7.762718245564498e-06,
"loss": 0.5819,
"step": 179500
},
{
"epoch": 2.54,
"learning_rate": 7.645065650148241e-06,
"loss": 0.5901,
"step": 180000
},
{
"epoch": 2.55,
"learning_rate": 7.527413054731988e-06,
"loss": 0.58,
"step": 180500
},
{
"epoch": 2.56,
"learning_rate": 7.409760459315733e-06,
"loss": 0.5963,
"step": 181000
},
{
"epoch": 2.56,
"learning_rate": 7.292107863899478e-06,
"loss": 0.604,
"step": 181500
},
{
"epoch": 2.57,
"learning_rate": 7.174455268483223e-06,
"loss": 0.5984,
"step": 182000
},
{
"epoch": 2.58,
"learning_rate": 7.056802673066969e-06,
"loss": 0.591,
"step": 182500
},
{
"epoch": 2.58,
"learning_rate": 6.939150077650713e-06,
"loss": 0.5932,
"step": 183000
},
{
"epoch": 2.59,
"learning_rate": 6.8214974822344585e-06,
"loss": 0.5996,
"step": 183500
},
{
"epoch": 2.6,
"learning_rate": 6.703844886818203e-06,
"loss": 0.5987,
"step": 184000
},
{
"epoch": 2.6,
"learning_rate": 6.586192291401949e-06,
"loss": 0.5859,
"step": 184500
},
{
"epoch": 2.61,
"learning_rate": 6.468539695985694e-06,
"loss": 0.594,
"step": 185000
},
{
"epoch": 2.62,
"learning_rate": 6.350887100569439e-06,
"loss": 0.5836,
"step": 185500
},
{
"epoch": 2.63,
"learning_rate": 6.233234505153184e-06,
"loss": 0.5789,
"step": 186000
},
{
"epoch": 2.63,
"learning_rate": 6.115581909736929e-06,
"loss": 0.595,
"step": 186500
},
{
"epoch": 2.64,
"learning_rate": 5.997929314320674e-06,
"loss": 0.5817,
"step": 187000
},
{
"epoch": 2.65,
"learning_rate": 5.880276718904419e-06,
"loss": 0.5926,
"step": 187500
},
{
"epoch": 2.65,
"learning_rate": 5.762624123488165e-06,
"loss": 0.6029,
"step": 188000
},
{
"epoch": 2.66,
"learning_rate": 5.644971528071909e-06,
"loss": 0.5888,
"step": 188500
},
{
"epoch": 2.67,
"learning_rate": 5.527318932655655e-06,
"loss": 0.5843,
"step": 189000
},
{
"epoch": 2.68,
"learning_rate": 5.4096663372394e-06,
"loss": 0.5794,
"step": 189500
},
{
"epoch": 2.68,
"learning_rate": 5.292013741823145e-06,
"loss": 0.5829,
"step": 190000
},
{
"epoch": 2.69,
"learning_rate": 5.1743611464068895e-06,
"loss": 0.5903,
"step": 190500
},
{
"epoch": 2.7,
"learning_rate": 5.056708550990636e-06,
"loss": 0.5742,
"step": 191000
},
{
"epoch": 2.7,
"learning_rate": 4.93905595557438e-06,
"loss": 0.5948,
"step": 191500
},
{
"epoch": 2.71,
"learning_rate": 4.821403360158125e-06,
"loss": 0.5807,
"step": 192000
},
{
"epoch": 2.72,
"learning_rate": 4.703750764741871e-06,
"loss": 0.5824,
"step": 192500
},
{
"epoch": 2.72,
"learning_rate": 4.586098169325616e-06,
"loss": 0.5818,
"step": 193000
},
{
"epoch": 2.73,
"learning_rate": 4.4684455739093605e-06,
"loss": 0.5871,
"step": 193500
},
{
"epoch": 2.74,
"learning_rate": 4.350792978493106e-06,
"loss": 0.5814,
"step": 194000
},
{
"epoch": 2.75,
"learning_rate": 4.233140383076851e-06,
"loss": 0.5851,
"step": 194500
},
{
"epoch": 2.75,
"learning_rate": 4.1154877876605964e-06,
"loss": 0.5794,
"step": 195000
},
{
"epoch": 2.76,
"learning_rate": 3.997835192244341e-06,
"loss": 0.5914,
"step": 195500
},
{
"epoch": 2.77,
"learning_rate": 3.880182596828086e-06,
"loss": 0.5792,
"step": 196000
},
{
"epoch": 2.77,
"learning_rate": 3.762530001411831e-06,
"loss": 0.5941,
"step": 196500
},
{
"epoch": 2.78,
"learning_rate": 3.644877405995577e-06,
"loss": 0.5673,
"step": 197000
},
{
"epoch": 2.79,
"learning_rate": 3.5272248105793217e-06,
"loss": 0.5818,
"step": 197500
},
{
"epoch": 2.8,
"learning_rate": 3.409572215163067e-06,
"loss": 0.5828,
"step": 198000
},
{
"epoch": 2.8,
"learning_rate": 3.291919619746812e-06,
"loss": 0.5798,
"step": 198500
},
{
"epoch": 2.81,
"learning_rate": 3.1742670243305572e-06,
"loss": 0.5714,
"step": 199000
},
{
"epoch": 2.82,
"learning_rate": 3.056614428914302e-06,
"loss": 0.5866,
"step": 199500
},
{
"epoch": 2.82,
"learning_rate": 2.938961833498047e-06,
"loss": 0.578,
"step": 200000
},
{
"epoch": 2.83,
"learning_rate": 2.8213092380817923e-06,
"loss": 0.5924,
"step": 200500
},
{
"epoch": 2.84,
"learning_rate": 2.703656642665537e-06,
"loss": 0.5854,
"step": 201000
},
{
"epoch": 2.84,
"learning_rate": 2.5860040472492825e-06,
"loss": 0.5865,
"step": 201500
},
{
"epoch": 2.85,
"learning_rate": 2.4683514518330274e-06,
"loss": 0.5933,
"step": 202000
},
{
"epoch": 2.86,
"learning_rate": 2.3506988564167727e-06,
"loss": 0.5748,
"step": 202500
},
{
"epoch": 2.87,
"learning_rate": 2.233046261000518e-06,
"loss": 0.5913,
"step": 203000
},
{
"epoch": 2.87,
"learning_rate": 2.115393665584263e-06,
"loss": 0.5811,
"step": 203500
},
{
"epoch": 2.88,
"learning_rate": 1.997741070168008e-06,
"loss": 0.5835,
"step": 204000
},
{
"epoch": 2.89,
"learning_rate": 1.880088474751753e-06,
"loss": 0.5836,
"step": 204500
},
{
"epoch": 2.89,
"learning_rate": 1.7624358793354984e-06,
"loss": 0.5737,
"step": 205000
},
{
"epoch": 2.9,
"learning_rate": 1.6447832839192435e-06,
"loss": 0.5813,
"step": 205500
},
{
"epoch": 2.91,
"learning_rate": 1.5271306885029886e-06,
"loss": 0.5837,
"step": 206000
},
{
"epoch": 2.92,
"learning_rate": 1.4094780930867337e-06,
"loss": 0.5816,
"step": 206500
},
{
"epoch": 2.92,
"learning_rate": 1.2918254976704786e-06,
"loss": 0.5871,
"step": 207000
},
{
"epoch": 2.93,
"learning_rate": 1.1741729022542237e-06,
"loss": 0.5657,
"step": 207500
},
{
"epoch": 2.94,
"learning_rate": 1.0565203068379688e-06,
"loss": 0.584,
"step": 208000
},
{
"epoch": 2.94,
"learning_rate": 9.38867711421714e-07,
"loss": 0.5672,
"step": 208500
},
{
"epoch": 2.95,
"learning_rate": 8.212151160054591e-07,
"loss": 0.5814,
"step": 209000
},
{
"epoch": 2.96,
"learning_rate": 7.035625205892043e-07,
"loss": 0.579,
"step": 209500
},
{
"epoch": 2.96,
"learning_rate": 5.859099251729494e-07,
"loss": 0.5969,
"step": 210000
},
{
"epoch": 2.97,
"learning_rate": 4.6825732975669443e-07,
"loss": 0.5796,
"step": 210500
},
{
"epoch": 2.98,
"learning_rate": 3.506047343404396e-07,
"loss": 0.5699,
"step": 211000
},
{
"epoch": 2.99,
"learning_rate": 2.3295213892418468e-07,
"loss": 0.5749,
"step": 211500
},
{
"epoch": 2.99,
"learning_rate": 1.152995435079298e-07,
"loss": 0.5678,
"step": 212000
},
{
"epoch": 3.0,
"step": 212490,
"total_flos": 4.485693256017408e+17,
"train_loss": 0.7508337134063662,
"train_runtime": 150510.9458,
"train_samples_per_second": 11.294,
"train_steps_per_second": 1.412
}
],
"logging_steps": 500,
"max_steps": 212490,
"num_train_epochs": 3,
"save_steps": 100000,
"total_flos": 4.485693256017408e+17,
"trial_name": null,
"trial_params": null
}