gpt2-4chan-mini / trainer_state.json
niizam
greentext dataset
08b5dcf
raw
history blame contribute delete
No virus
27.2 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9983934941049403,
"global_step": 110000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.9924364129234474e-05,
"loss": 4.7552,
"step": 500
},
{
"epoch": 0.01,
"learning_rate": 4.984872825846895e-05,
"loss": 4.6634,
"step": 1000
},
{
"epoch": 0.01,
"learning_rate": 4.9773092387703424e-05,
"loss": 4.583,
"step": 1500
},
{
"epoch": 0.02,
"learning_rate": 4.96974565169379e-05,
"loss": 4.5872,
"step": 2000
},
{
"epoch": 0.02,
"learning_rate": 4.962182064617237e-05,
"loss": 4.5253,
"step": 2500
},
{
"epoch": 0.03,
"learning_rate": 4.9546184775406845e-05,
"loss": 4.4986,
"step": 3000
},
{
"epoch": 0.03,
"learning_rate": 4.947054890464132e-05,
"loss": 4.5247,
"step": 3500
},
{
"epoch": 0.04,
"learning_rate": 4.9394913033875794e-05,
"loss": 4.4397,
"step": 4000
},
{
"epoch": 0.04,
"learning_rate": 4.931927716311027e-05,
"loss": 4.5077,
"step": 4500
},
{
"epoch": 0.05,
"learning_rate": 4.9243641292344744e-05,
"loss": 4.4674,
"step": 5000
},
{
"epoch": 0.05,
"learning_rate": 4.9168005421579216e-05,
"loss": 4.4637,
"step": 5500
},
{
"epoch": 0.05,
"learning_rate": 4.9092369550813694e-05,
"loss": 4.4311,
"step": 6000
},
{
"epoch": 0.06,
"learning_rate": 4.9016733680048165e-05,
"loss": 4.4189,
"step": 6500
},
{
"epoch": 0.06,
"learning_rate": 4.8941097809282644e-05,
"loss": 4.3811,
"step": 7000
},
{
"epoch": 0.07,
"learning_rate": 4.8865461938517115e-05,
"loss": 4.4015,
"step": 7500
},
{
"epoch": 0.07,
"learning_rate": 4.8789826067751587e-05,
"loss": 4.3717,
"step": 8000
},
{
"epoch": 0.08,
"learning_rate": 4.8714190196986065e-05,
"loss": 4.3931,
"step": 8500
},
{
"epoch": 0.08,
"learning_rate": 4.8638554326220536e-05,
"loss": 4.3651,
"step": 9000
},
{
"epoch": 0.09,
"learning_rate": 4.8562918455455014e-05,
"loss": 4.3974,
"step": 9500
},
{
"epoch": 0.09,
"learning_rate": 4.8487282584689486e-05,
"loss": 4.3297,
"step": 10000
},
{
"epoch": 0.1,
"learning_rate": 4.8411646713923964e-05,
"loss": 4.3414,
"step": 10500
},
{
"epoch": 0.1,
"learning_rate": 4.8336010843158436e-05,
"loss": 4.3682,
"step": 11000
},
{
"epoch": 0.1,
"learning_rate": 4.826037497239291e-05,
"loss": 4.3614,
"step": 11500
},
{
"epoch": 0.11,
"learning_rate": 4.8184739101627385e-05,
"loss": 4.368,
"step": 12000
},
{
"epoch": 0.11,
"learning_rate": 4.810910323086186e-05,
"loss": 4.3026,
"step": 12500
},
{
"epoch": 0.12,
"learning_rate": 4.8033467360096335e-05,
"loss": 4.3342,
"step": 13000
},
{
"epoch": 0.12,
"learning_rate": 4.7957831489330807e-05,
"loss": 4.3359,
"step": 13500
},
{
"epoch": 0.13,
"learning_rate": 4.788219561856528e-05,
"loss": 4.3332,
"step": 14000
},
{
"epoch": 0.13,
"learning_rate": 4.7806559747799756e-05,
"loss": 4.2581,
"step": 14500
},
{
"epoch": 0.14,
"learning_rate": 4.773092387703423e-05,
"loss": 4.3509,
"step": 15000
},
{
"epoch": 0.14,
"learning_rate": 4.7655288006268706e-05,
"loss": 4.2819,
"step": 15500
},
{
"epoch": 0.15,
"learning_rate": 4.757965213550318e-05,
"loss": 4.2966,
"step": 16000
},
{
"epoch": 0.15,
"learning_rate": 4.750401626473765e-05,
"loss": 4.2967,
"step": 16500
},
{
"epoch": 0.15,
"learning_rate": 4.742838039397213e-05,
"loss": 4.2915,
"step": 17000
},
{
"epoch": 0.16,
"learning_rate": 4.73527445232066e-05,
"loss": 4.2473,
"step": 17500
},
{
"epoch": 0.16,
"learning_rate": 4.727710865244108e-05,
"loss": 4.2357,
"step": 18000
},
{
"epoch": 0.17,
"learning_rate": 4.720147278167555e-05,
"loss": 4.2754,
"step": 18500
},
{
"epoch": 0.17,
"learning_rate": 4.712583691091002e-05,
"loss": 4.2699,
"step": 19000
},
{
"epoch": 0.18,
"learning_rate": 4.70502010401445e-05,
"loss": 4.2866,
"step": 19500
},
{
"epoch": 0.18,
"learning_rate": 4.697456516937897e-05,
"loss": 4.2778,
"step": 20000
},
{
"epoch": 0.19,
"learning_rate": 4.689892929861345e-05,
"loss": 4.2414,
"step": 20500
},
{
"epoch": 0.19,
"learning_rate": 4.682329342784792e-05,
"loss": 4.2963,
"step": 21000
},
{
"epoch": 0.2,
"learning_rate": 4.674765755708239e-05,
"loss": 4.3053,
"step": 21500
},
{
"epoch": 0.2,
"learning_rate": 4.667202168631687e-05,
"loss": 4.2914,
"step": 22000
},
{
"epoch": 0.2,
"learning_rate": 4.659638581555134e-05,
"loss": 4.2527,
"step": 22500
},
{
"epoch": 0.21,
"learning_rate": 4.652074994478582e-05,
"loss": 4.251,
"step": 23000
},
{
"epoch": 0.21,
"learning_rate": 4.644511407402029e-05,
"loss": 4.2694,
"step": 23500
},
{
"epoch": 0.22,
"learning_rate": 4.636947820325476e-05,
"loss": 4.2306,
"step": 24000
},
{
"epoch": 0.22,
"learning_rate": 4.629384233248924e-05,
"loss": 4.2451,
"step": 24500
},
{
"epoch": 0.23,
"learning_rate": 4.621820646172371e-05,
"loss": 4.232,
"step": 25000
},
{
"epoch": 0.23,
"learning_rate": 4.614257059095819e-05,
"loss": 4.2506,
"step": 25500
},
{
"epoch": 0.24,
"learning_rate": 4.606693472019266e-05,
"loss": 4.2425,
"step": 26000
},
{
"epoch": 0.24,
"learning_rate": 4.599129884942713e-05,
"loss": 4.2495,
"step": 26500
},
{
"epoch": 0.25,
"learning_rate": 4.591566297866161e-05,
"loss": 4.2475,
"step": 27000
},
{
"epoch": 0.25,
"learning_rate": 4.584002710789608e-05,
"loss": 4.2516,
"step": 27500
},
{
"epoch": 0.25,
"learning_rate": 4.576439123713056e-05,
"loss": 4.2478,
"step": 28000
},
{
"epoch": 0.26,
"learning_rate": 4.568875536636503e-05,
"loss": 4.2214,
"step": 28500
},
{
"epoch": 0.26,
"learning_rate": 4.56131194955995e-05,
"loss": 4.244,
"step": 29000
},
{
"epoch": 0.27,
"learning_rate": 4.553748362483398e-05,
"loss": 4.2125,
"step": 29500
},
{
"epoch": 0.27,
"learning_rate": 4.546184775406845e-05,
"loss": 4.251,
"step": 30000
},
{
"epoch": 0.28,
"learning_rate": 4.538621188330293e-05,
"loss": 4.1668,
"step": 30500
},
{
"epoch": 0.28,
"learning_rate": 4.53105760125374e-05,
"loss": 4.2469,
"step": 31000
},
{
"epoch": 0.29,
"learning_rate": 4.523494014177188e-05,
"loss": 4.2189,
"step": 31500
},
{
"epoch": 0.29,
"learning_rate": 4.515930427100635e-05,
"loss": 4.2727,
"step": 32000
},
{
"epoch": 0.29,
"learning_rate": 4.5083668400240824e-05,
"loss": 4.2318,
"step": 32500
},
{
"epoch": 0.3,
"learning_rate": 4.50080325294753e-05,
"loss": 4.2083,
"step": 33000
},
{
"epoch": 0.3,
"learning_rate": 4.4932396658709774e-05,
"loss": 4.1744,
"step": 33500
},
{
"epoch": 0.31,
"learning_rate": 4.485676078794425e-05,
"loss": 4.1938,
"step": 34000
},
{
"epoch": 0.31,
"learning_rate": 4.478112491717872e-05,
"loss": 4.1745,
"step": 34500
},
{
"epoch": 0.32,
"learning_rate": 4.4705489046413195e-05,
"loss": 4.1725,
"step": 35000
},
{
"epoch": 0.32,
"learning_rate": 4.462985317564767e-05,
"loss": 4.2081,
"step": 35500
},
{
"epoch": 0.33,
"learning_rate": 4.4554217304882145e-05,
"loss": 4.1804,
"step": 36000
},
{
"epoch": 0.33,
"learning_rate": 4.447858143411662e-05,
"loss": 4.1977,
"step": 36500
},
{
"epoch": 0.34,
"learning_rate": 4.4402945563351094e-05,
"loss": 4.1939,
"step": 37000
},
{
"epoch": 0.34,
"learning_rate": 4.4327309692585566e-05,
"loss": 4.2137,
"step": 37500
},
{
"epoch": 0.34,
"learning_rate": 4.4251673821820044e-05,
"loss": 4.1899,
"step": 38000
},
{
"epoch": 0.35,
"learning_rate": 4.4176037951054515e-05,
"loss": 4.1409,
"step": 38500
},
{
"epoch": 0.35,
"learning_rate": 4.4100402080288994e-05,
"loss": 4.155,
"step": 39000
},
{
"epoch": 0.36,
"learning_rate": 4.4024766209523465e-05,
"loss": 4.2214,
"step": 39500
},
{
"epoch": 0.36,
"learning_rate": 4.3949130338757937e-05,
"loss": 4.1553,
"step": 40000
},
{
"epoch": 0.37,
"learning_rate": 4.3873494467992415e-05,
"loss": 4.1859,
"step": 40500
},
{
"epoch": 0.37,
"learning_rate": 4.3797858597226886e-05,
"loss": 4.2127,
"step": 41000
},
{
"epoch": 0.38,
"learning_rate": 4.3722222726461365e-05,
"loss": 4.1758,
"step": 41500
},
{
"epoch": 0.38,
"learning_rate": 4.3646586855695836e-05,
"loss": 4.2035,
"step": 42000
},
{
"epoch": 0.39,
"learning_rate": 4.357095098493031e-05,
"loss": 4.1486,
"step": 42500
},
{
"epoch": 0.39,
"learning_rate": 4.3495315114164786e-05,
"loss": 4.2252,
"step": 43000
},
{
"epoch": 0.39,
"learning_rate": 4.341967924339926e-05,
"loss": 4.1585,
"step": 43500
},
{
"epoch": 0.4,
"learning_rate": 4.3344043372633735e-05,
"loss": 4.1804,
"step": 44000
},
{
"epoch": 0.4,
"learning_rate": 4.326840750186821e-05,
"loss": 4.2005,
"step": 44500
},
{
"epoch": 0.41,
"learning_rate": 4.319277163110268e-05,
"loss": 4.1257,
"step": 45000
},
{
"epoch": 0.41,
"learning_rate": 4.3117135760337157e-05,
"loss": 4.1687,
"step": 45500
},
{
"epoch": 0.42,
"learning_rate": 4.304149988957163e-05,
"loss": 4.1845,
"step": 46000
},
{
"epoch": 0.42,
"learning_rate": 4.2965864018806106e-05,
"loss": 4.1383,
"step": 46500
},
{
"epoch": 0.43,
"learning_rate": 4.289022814804058e-05,
"loss": 4.1346,
"step": 47000
},
{
"epoch": 0.43,
"learning_rate": 4.281459227727505e-05,
"loss": 4.1611,
"step": 47500
},
{
"epoch": 0.44,
"learning_rate": 4.273895640650953e-05,
"loss": 4.1542,
"step": 48000
},
{
"epoch": 0.44,
"learning_rate": 4.2663320535744e-05,
"loss": 4.2022,
"step": 48500
},
{
"epoch": 0.44,
"learning_rate": 4.258768466497848e-05,
"loss": 4.1973,
"step": 49000
},
{
"epoch": 0.45,
"learning_rate": 4.251204879421295e-05,
"loss": 4.1386,
"step": 49500
},
{
"epoch": 0.45,
"learning_rate": 4.243641292344743e-05,
"loss": 4.1542,
"step": 50000
},
{
"epoch": 0.46,
"learning_rate": 4.23607770526819e-05,
"loss": 4.1301,
"step": 50500
},
{
"epoch": 0.46,
"learning_rate": 4.228514118191637e-05,
"loss": 4.1239,
"step": 51000
},
{
"epoch": 0.47,
"learning_rate": 4.220950531115085e-05,
"loss": 4.1453,
"step": 51500
},
{
"epoch": 0.47,
"learning_rate": 4.213386944038532e-05,
"loss": 4.1403,
"step": 52000
},
{
"epoch": 0.48,
"learning_rate": 4.20582335696198e-05,
"loss": 4.1665,
"step": 52500
},
{
"epoch": 0.48,
"learning_rate": 4.198259769885427e-05,
"loss": 4.1306,
"step": 53000
},
{
"epoch": 0.49,
"learning_rate": 4.190696182808874e-05,
"loss": 4.1622,
"step": 53500
},
{
"epoch": 0.49,
"learning_rate": 4.183132595732322e-05,
"loss": 4.1524,
"step": 54000
},
{
"epoch": 0.49,
"learning_rate": 4.175569008655769e-05,
"loss": 4.1656,
"step": 54500
},
{
"epoch": 0.5,
"learning_rate": 4.168005421579217e-05,
"loss": 4.092,
"step": 55000
},
{
"epoch": 0.5,
"learning_rate": 4.160441834502664e-05,
"loss": 4.1483,
"step": 55500
},
{
"epoch": 0.51,
"learning_rate": 4.152878247426111e-05,
"loss": 4.1474,
"step": 56000
},
{
"epoch": 0.51,
"learning_rate": 4.145314660349559e-05,
"loss": 4.1685,
"step": 56500
},
{
"epoch": 0.52,
"learning_rate": 4.137751073273006e-05,
"loss": 4.1404,
"step": 57000
},
{
"epoch": 0.52,
"learning_rate": 4.130187486196454e-05,
"loss": 4.0895,
"step": 57500
},
{
"epoch": 0.53,
"learning_rate": 4.122623899119901e-05,
"loss": 4.1132,
"step": 58000
},
{
"epoch": 0.53,
"learning_rate": 4.115060312043348e-05,
"loss": 4.1438,
"step": 58500
},
{
"epoch": 0.54,
"learning_rate": 4.107496724966796e-05,
"loss": 4.1197,
"step": 59000
},
{
"epoch": 0.54,
"learning_rate": 4.099933137890243e-05,
"loss": 4.1568,
"step": 59500
},
{
"epoch": 0.54,
"learning_rate": 4.092369550813691e-05,
"loss": 4.1546,
"step": 60000
},
{
"epoch": 0.55,
"learning_rate": 4.084805963737138e-05,
"loss": 4.1391,
"step": 60500
},
{
"epoch": 0.55,
"learning_rate": 4.077242376660585e-05,
"loss": 4.1197,
"step": 61000
},
{
"epoch": 0.56,
"learning_rate": 4.069678789584033e-05,
"loss": 4.1323,
"step": 61500
},
{
"epoch": 0.56,
"learning_rate": 4.06211520250748e-05,
"loss": 4.1305,
"step": 62000
},
{
"epoch": 0.57,
"learning_rate": 4.054551615430928e-05,
"loss": 4.0884,
"step": 62500
},
{
"epoch": 0.57,
"learning_rate": 4.046988028354375e-05,
"loss": 4.1144,
"step": 63000
},
{
"epoch": 0.58,
"learning_rate": 4.0394244412778224e-05,
"loss": 4.1014,
"step": 63500
},
{
"epoch": 0.58,
"learning_rate": 4.03186085420127e-05,
"loss": 4.0915,
"step": 64000
},
{
"epoch": 0.59,
"learning_rate": 4.0242972671247174e-05,
"loss": 4.158,
"step": 64500
},
{
"epoch": 0.59,
"learning_rate": 4.016733680048165e-05,
"loss": 4.13,
"step": 65000
},
{
"epoch": 0.59,
"learning_rate": 4.0091700929716124e-05,
"loss": 4.1329,
"step": 65500
},
{
"epoch": 0.6,
"learning_rate": 4.0016065058950595e-05,
"loss": 4.0738,
"step": 66000
},
{
"epoch": 0.6,
"learning_rate": 3.994042918818507e-05,
"loss": 4.1312,
"step": 66500
},
{
"epoch": 0.61,
"learning_rate": 3.9864793317419545e-05,
"loss": 4.1058,
"step": 67000
},
{
"epoch": 0.61,
"learning_rate": 3.978915744665402e-05,
"loss": 4.0855,
"step": 67500
},
{
"epoch": 0.62,
"learning_rate": 3.9713521575888495e-05,
"loss": 4.1531,
"step": 68000
},
{
"epoch": 0.62,
"learning_rate": 3.9637885705122966e-05,
"loss": 4.1676,
"step": 68500
},
{
"epoch": 0.63,
"learning_rate": 3.9562249834357444e-05,
"loss": 4.0996,
"step": 69000
},
{
"epoch": 0.63,
"learning_rate": 3.9486613963591916e-05,
"loss": 4.0781,
"step": 69500
},
{
"epoch": 0.64,
"learning_rate": 3.9410978092826394e-05,
"loss": 4.0646,
"step": 70000
},
{
"epoch": 0.64,
"learning_rate": 3.9335342222060865e-05,
"loss": 4.0895,
"step": 70500
},
{
"epoch": 0.64,
"learning_rate": 3.9259706351295344e-05,
"loss": 4.1257,
"step": 71000
},
{
"epoch": 0.65,
"learning_rate": 3.9184070480529815e-05,
"loss": 4.0724,
"step": 71500
},
{
"epoch": 0.65,
"learning_rate": 3.910843460976429e-05,
"loss": 4.082,
"step": 72000
},
{
"epoch": 0.66,
"learning_rate": 3.9032798738998765e-05,
"loss": 4.1544,
"step": 72500
},
{
"epoch": 0.66,
"learning_rate": 3.8957162868233236e-05,
"loss": 4.124,
"step": 73000
},
{
"epoch": 0.67,
"learning_rate": 3.8881526997467715e-05,
"loss": 4.0763,
"step": 73500
},
{
"epoch": 0.67,
"learning_rate": 3.8805891126702186e-05,
"loss": 4.1284,
"step": 74000
},
{
"epoch": 0.68,
"learning_rate": 3.873025525593666e-05,
"loss": 4.0564,
"step": 74500
},
{
"epoch": 0.68,
"learning_rate": 3.8654619385171136e-05,
"loss": 4.1,
"step": 75000
},
{
"epoch": 0.69,
"learning_rate": 3.857898351440561e-05,
"loss": 4.1089,
"step": 75500
},
{
"epoch": 0.69,
"learning_rate": 3.8503347643640085e-05,
"loss": 4.092,
"step": 76000
},
{
"epoch": 0.69,
"learning_rate": 3.842771177287456e-05,
"loss": 4.127,
"step": 76500
},
{
"epoch": 0.7,
"learning_rate": 3.835207590210903e-05,
"loss": 4.1307,
"step": 77000
},
{
"epoch": 0.7,
"learning_rate": 3.827644003134351e-05,
"loss": 4.0732,
"step": 77500
},
{
"epoch": 0.71,
"learning_rate": 3.820080416057798e-05,
"loss": 4.0956,
"step": 78000
},
{
"epoch": 0.71,
"learning_rate": 3.8125168289812456e-05,
"loss": 4.0683,
"step": 78500
},
{
"epoch": 0.72,
"learning_rate": 3.804953241904693e-05,
"loss": 4.0822,
"step": 79000
},
{
"epoch": 0.72,
"learning_rate": 3.79738965482814e-05,
"loss": 4.1471,
"step": 79500
},
{
"epoch": 0.73,
"learning_rate": 3.789826067751588e-05,
"loss": 4.127,
"step": 80000
},
{
"epoch": 0.73,
"learning_rate": 3.782262480675035e-05,
"loss": 4.1113,
"step": 80500
},
{
"epoch": 0.74,
"learning_rate": 3.774698893598483e-05,
"loss": 4.0932,
"step": 81000
},
{
"epoch": 0.74,
"learning_rate": 3.76713530652193e-05,
"loss": 4.1176,
"step": 81500
},
{
"epoch": 0.74,
"learning_rate": 3.759571719445377e-05,
"loss": 4.1078,
"step": 82000
},
{
"epoch": 0.75,
"learning_rate": 3.752008132368825e-05,
"loss": 4.0988,
"step": 82500
},
{
"epoch": 0.75,
"learning_rate": 3.744444545292272e-05,
"loss": 4.1075,
"step": 83000
},
{
"epoch": 0.76,
"learning_rate": 3.73688095821572e-05,
"loss": 4.0456,
"step": 83500
},
{
"epoch": 0.76,
"learning_rate": 3.729317371139167e-05,
"loss": 4.1451,
"step": 84000
},
{
"epoch": 0.77,
"learning_rate": 3.721753784062614e-05,
"loss": 4.116,
"step": 84500
},
{
"epoch": 0.77,
"learning_rate": 3.714190196986062e-05,
"loss": 4.0827,
"step": 85000
},
{
"epoch": 0.78,
"learning_rate": 3.706626609909509e-05,
"loss": 4.1093,
"step": 85500
},
{
"epoch": 0.78,
"learning_rate": 3.699063022832957e-05,
"loss": 4.0878,
"step": 86000
},
{
"epoch": 0.79,
"learning_rate": 3.691499435756404e-05,
"loss": 4.1075,
"step": 86500
},
{
"epoch": 0.79,
"learning_rate": 3.683935848679851e-05,
"loss": 4.0942,
"step": 87000
},
{
"epoch": 0.79,
"learning_rate": 3.676372261603299e-05,
"loss": 4.1082,
"step": 87500
},
{
"epoch": 0.8,
"learning_rate": 3.668808674526746e-05,
"loss": 4.0797,
"step": 88000
},
{
"epoch": 0.8,
"learning_rate": 3.661245087450194e-05,
"loss": 4.0613,
"step": 88500
},
{
"epoch": 0.81,
"learning_rate": 3.653681500373641e-05,
"loss": 4.1392,
"step": 89000
},
{
"epoch": 0.81,
"learning_rate": 3.646117913297089e-05,
"loss": 4.064,
"step": 89500
},
{
"epoch": 0.82,
"learning_rate": 3.638554326220536e-05,
"loss": 4.0528,
"step": 90000
},
{
"epoch": 0.82,
"learning_rate": 3.630990739143983e-05,
"loss": 4.0703,
"step": 90500
},
{
"epoch": 0.83,
"learning_rate": 3.623427152067431e-05,
"loss": 4.0334,
"step": 91000
},
{
"epoch": 0.83,
"learning_rate": 3.615863564990878e-05,
"loss": 4.0341,
"step": 91500
},
{
"epoch": 0.84,
"learning_rate": 3.608299977914326e-05,
"loss": 4.079,
"step": 92000
},
{
"epoch": 0.84,
"learning_rate": 3.600736390837773e-05,
"loss": 4.0658,
"step": 92500
},
{
"epoch": 0.84,
"learning_rate": 3.5931728037612203e-05,
"loss": 4.0723,
"step": 93000
},
{
"epoch": 0.85,
"learning_rate": 3.585609216684668e-05,
"loss": 4.0875,
"step": 93500
},
{
"epoch": 0.85,
"learning_rate": 3.578045629608115e-05,
"loss": 4.0256,
"step": 94000
},
{
"epoch": 0.86,
"learning_rate": 3.570482042531563e-05,
"loss": 4.1114,
"step": 94500
},
{
"epoch": 0.86,
"learning_rate": 3.56291845545501e-05,
"loss": 4.0955,
"step": 95000
},
{
"epoch": 0.87,
"learning_rate": 3.5553548683784574e-05,
"loss": 4.1109,
"step": 95500
},
{
"epoch": 0.87,
"learning_rate": 3.547791281301905e-05,
"loss": 4.0938,
"step": 96000
},
{
"epoch": 0.88,
"learning_rate": 3.5402276942253524e-05,
"loss": 4.0839,
"step": 96500
},
{
"epoch": 0.88,
"learning_rate": 3.5326641071488e-05,
"loss": 4.0671,
"step": 97000
},
{
"epoch": 0.88,
"learning_rate": 3.5251005200722474e-05,
"loss": 4.0446,
"step": 97500
},
{
"epoch": 0.89,
"learning_rate": 3.5175369329956945e-05,
"loss": 4.0323,
"step": 98000
},
{
"epoch": 0.89,
"learning_rate": 3.5099733459191423e-05,
"loss": 4.0442,
"step": 98500
},
{
"epoch": 0.9,
"learning_rate": 3.5024097588425895e-05,
"loss": 4.0989,
"step": 99000
},
{
"epoch": 0.9,
"learning_rate": 3.494846171766037e-05,
"loss": 4.0851,
"step": 99500
},
{
"epoch": 0.91,
"learning_rate": 3.4872825846894845e-05,
"loss": 4.0733,
"step": 100000
},
{
"epoch": 0.91,
"learning_rate": 3.4797189976129316e-05,
"loss": 4.0693,
"step": 100500
},
{
"epoch": 0.92,
"learning_rate": 3.4721554105363794e-05,
"loss": 4.0285,
"step": 101000
},
{
"epoch": 0.92,
"learning_rate": 3.4645918234598266e-05,
"loss": 4.1077,
"step": 101500
},
{
"epoch": 0.93,
"learning_rate": 3.4570282363832744e-05,
"loss": 4.0844,
"step": 102000
},
{
"epoch": 0.93,
"learning_rate": 3.4494646493067215e-05,
"loss": 4.0826,
"step": 102500
},
{
"epoch": 0.93,
"learning_rate": 3.441901062230169e-05,
"loss": 4.0967,
"step": 103000
},
{
"epoch": 0.94,
"learning_rate": 3.4343374751536165e-05,
"loss": 4.0536,
"step": 103500
},
{
"epoch": 0.94,
"learning_rate": 3.426773888077064e-05,
"loss": 4.1287,
"step": 104000
},
{
"epoch": 0.95,
"learning_rate": 3.4192103010005115e-05,
"loss": 4.0883,
"step": 104500
},
{
"epoch": 0.95,
"learning_rate": 3.4116467139239586e-05,
"loss": 4.0862,
"step": 105000
},
{
"epoch": 0.96,
"learning_rate": 3.404083126847406e-05,
"loss": 4.0683,
"step": 105500
},
{
"epoch": 0.96,
"learning_rate": 3.3965195397708536e-05,
"loss": 4.0465,
"step": 106000
},
{
"epoch": 0.97,
"learning_rate": 3.388955952694301e-05,
"loss": 4.0712,
"step": 106500
},
{
"epoch": 0.97,
"learning_rate": 3.3813923656177486e-05,
"loss": 4.0737,
"step": 107000
},
{
"epoch": 0.98,
"learning_rate": 3.373828778541196e-05,
"loss": 4.0399,
"step": 107500
},
{
"epoch": 0.98,
"learning_rate": 3.366265191464643e-05,
"loss": 4.0716,
"step": 108000
},
{
"epoch": 0.98,
"learning_rate": 3.358701604388091e-05,
"loss": 4.0129,
"step": 108500
},
{
"epoch": 0.99,
"learning_rate": 3.351138017311538e-05,
"loss": 4.0711,
"step": 109000
},
{
"epoch": 0.99,
"learning_rate": 3.343574430234986e-05,
"loss": 4.0583,
"step": 109500
},
{
"epoch": 1.0,
"learning_rate": 3.336010843158433e-05,
"loss": 4.0821,
"step": 110000
}
],
"max_steps": 330531,
"num_train_epochs": 3,
"total_flos": 7185530880000000.0,
"trial_name": null,
"trial_params": null
}