enko_mbartLarge_sup_run1 / trainer_state.json
yesj1234's picture
Upload folder using huggingface_hub
da68fa1 verified
{
"best_metric": 0.6636257171630859,
"best_model_checkpoint": "./enko_mbartLarge_100p_run1/checkpoint-190759",
"epoch": 6.0,
"eval_steps": 500,
"global_step": 381519,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 9.940000000000001e-06,
"loss": 1.6525,
"step": 500
},
{
"epoch": 0.02,
"learning_rate": 1.994e-05,
"loss": 1.3306,
"step": 1000
},
{
"epoch": 0.02,
"learning_rate": 2.994e-05,
"loss": 1.3005,
"step": 1500
},
{
"epoch": 0.03,
"learning_rate": 3.994e-05,
"loss": 1.2871,
"step": 2000
},
{
"epoch": 0.04,
"learning_rate": 4.9940000000000006e-05,
"loss": 1.306,
"step": 2500
},
{
"epoch": 0.05,
"learning_rate": 4.9973930136971904e-05,
"loss": 1.2876,
"step": 3000
},
{
"epoch": 0.06,
"learning_rate": 4.994770259332065e-05,
"loss": 1.2609,
"step": 3500
},
{
"epoch": 0.06,
"learning_rate": 4.992142248946168e-05,
"loss": 1.244,
"step": 4000
},
{
"epoch": 0.07,
"learning_rate": 4.9895142385602714e-05,
"loss": 1.2051,
"step": 4500
},
{
"epoch": 0.08,
"learning_rate": 4.986886228174374e-05,
"loss": 1.1997,
"step": 5000
},
{
"epoch": 0.09,
"learning_rate": 4.9842582177884766e-05,
"loss": 1.1854,
"step": 5500
},
{
"epoch": 0.09,
"learning_rate": 4.98163020740258e-05,
"loss": 1.1666,
"step": 6000
},
{
"epoch": 0.1,
"learning_rate": 4.9790021970166825e-05,
"loss": 1.1596,
"step": 6500
},
{
"epoch": 0.11,
"learning_rate": 4.976374186630786e-05,
"loss": 1.1559,
"step": 7000
},
{
"epoch": 0.12,
"learning_rate": 4.973746176244889e-05,
"loss": 1.1555,
"step": 7500
},
{
"epoch": 0.13,
"learning_rate": 4.9711181658589916e-05,
"loss": 1.1458,
"step": 8000
},
{
"epoch": 0.13,
"learning_rate": 4.968495411493866e-05,
"loss": 1.1098,
"step": 8500
},
{
"epoch": 0.14,
"learning_rate": 4.9658674011079694e-05,
"loss": 1.1169,
"step": 9000
},
{
"epoch": 0.15,
"learning_rate": 4.963239390722073e-05,
"loss": 1.114,
"step": 9500
},
{
"epoch": 0.16,
"learning_rate": 4.960616636356947e-05,
"loss": 1.1099,
"step": 10000
},
{
"epoch": 0.17,
"learning_rate": 4.95798862597105e-05,
"loss": 1.0846,
"step": 10500
},
{
"epoch": 0.17,
"learning_rate": 4.955360615585153e-05,
"loss": 1.1021,
"step": 11000
},
{
"epoch": 0.18,
"learning_rate": 4.952732605199256e-05,
"loss": 1.0739,
"step": 11500
},
{
"epoch": 0.19,
"learning_rate": 4.950104594813359e-05,
"loss": 1.0657,
"step": 12000
},
{
"epoch": 0.2,
"learning_rate": 4.9474765844274615e-05,
"loss": 1.0641,
"step": 12500
},
{
"epoch": 0.2,
"learning_rate": 4.944848574041565e-05,
"loss": 1.0696,
"step": 13000
},
{
"epoch": 0.21,
"learning_rate": 4.94222581967644e-05,
"loss": 1.0457,
"step": 13500
},
{
"epoch": 0.22,
"learning_rate": 4.9396030653113144e-05,
"loss": 1.0573,
"step": 14000
},
{
"epoch": 0.23,
"learning_rate": 4.936975054925417e-05,
"loss": 1.0302,
"step": 14500
},
{
"epoch": 0.24,
"learning_rate": 4.93434704453952e-05,
"loss": 1.0347,
"step": 15000
},
{
"epoch": 0.24,
"learning_rate": 4.9317190341536236e-05,
"loss": 1.0304,
"step": 15500
},
{
"epoch": 0.25,
"learning_rate": 4.929096279788498e-05,
"loss": 1.0454,
"step": 16000
},
{
"epoch": 0.26,
"learning_rate": 4.926468269402601e-05,
"loss": 1.0406,
"step": 16500
},
{
"epoch": 0.27,
"learning_rate": 4.923840259016704e-05,
"loss": 1.0306,
"step": 17000
},
{
"epoch": 0.28,
"learning_rate": 4.921212248630807e-05,
"loss": 1.0306,
"step": 17500
},
{
"epoch": 0.28,
"learning_rate": 4.91858423824491e-05,
"loss": 1.0237,
"step": 18000
},
{
"epoch": 0.29,
"learning_rate": 4.9159562278590124e-05,
"loss": 1.0113,
"step": 18500
},
{
"epoch": 0.3,
"learning_rate": 4.913328217473116e-05,
"loss": 1.0101,
"step": 19000
},
{
"epoch": 0.31,
"learning_rate": 4.910700207087219e-05,
"loss": 0.9928,
"step": 19500
},
{
"epoch": 0.31,
"learning_rate": 4.9080721967013216e-05,
"loss": 0.9942,
"step": 20000
},
{
"epoch": 0.32,
"learning_rate": 4.905444186315424e-05,
"loss": 1.0043,
"step": 20500
},
{
"epoch": 0.33,
"learning_rate": 4.9028161759295275e-05,
"loss": 1.0028,
"step": 21000
},
{
"epoch": 0.34,
"learning_rate": 4.900188165543631e-05,
"loss": 0.9883,
"step": 21500
},
{
"epoch": 0.35,
"learning_rate": 4.897565411178505e-05,
"loss": 1.0023,
"step": 22000
},
{
"epoch": 0.35,
"learning_rate": 4.89494265681338e-05,
"loss": 0.9935,
"step": 22500
},
{
"epoch": 0.36,
"learning_rate": 4.892314646427483e-05,
"loss": 0.9928,
"step": 23000
},
{
"epoch": 0.37,
"learning_rate": 4.889686636041586e-05,
"loss": 0.9948,
"step": 23500
},
{
"epoch": 0.38,
"learning_rate": 4.887058625655689e-05,
"loss": 0.974,
"step": 24000
},
{
"epoch": 0.39,
"learning_rate": 4.8844306152697915e-05,
"loss": 0.9837,
"step": 24500
},
{
"epoch": 0.39,
"learning_rate": 4.881802604883895e-05,
"loss": 0.9751,
"step": 25000
},
{
"epoch": 0.4,
"learning_rate": 4.879174594497998e-05,
"loss": 0.9702,
"step": 25500
},
{
"epoch": 0.41,
"learning_rate": 4.8765465841121006e-05,
"loss": 0.9717,
"step": 26000
},
{
"epoch": 0.42,
"learning_rate": 4.873918573726204e-05,
"loss": 0.9724,
"step": 26500
},
{
"epoch": 0.42,
"learning_rate": 4.8712905633403065e-05,
"loss": 0.9692,
"step": 27000
},
{
"epoch": 0.43,
"learning_rate": 4.868667808975182e-05,
"loss": 0.9535,
"step": 27500
},
{
"epoch": 0.44,
"learning_rate": 4.866039798589284e-05,
"loss": 0.9641,
"step": 28000
},
{
"epoch": 0.45,
"learning_rate": 4.863417044224159e-05,
"loss": 0.9774,
"step": 28500
},
{
"epoch": 0.46,
"learning_rate": 4.860789033838262e-05,
"loss": 0.9671,
"step": 29000
},
{
"epoch": 0.46,
"learning_rate": 4.858161023452365e-05,
"loss": 0.9458,
"step": 29500
},
{
"epoch": 0.47,
"learning_rate": 4.855533013066468e-05,
"loss": 0.9606,
"step": 30000
},
{
"epoch": 0.48,
"learning_rate": 4.8529050026805705e-05,
"loss": 0.9557,
"step": 30500
},
{
"epoch": 0.49,
"learning_rate": 4.850276992294674e-05,
"loss": 0.9664,
"step": 31000
},
{
"epoch": 0.5,
"learning_rate": 4.8476489819087764e-05,
"loss": 0.9427,
"step": 31500
},
{
"epoch": 0.5,
"learning_rate": 4.84502097152288e-05,
"loss": 0.957,
"step": 32000
},
{
"epoch": 0.51,
"learning_rate": 4.842392961136983e-05,
"loss": 0.9466,
"step": 32500
},
{
"epoch": 0.52,
"learning_rate": 4.8397649507510856e-05,
"loss": 0.9262,
"step": 33000
},
{
"epoch": 0.53,
"learning_rate": 4.837136940365188e-05,
"loss": 0.9524,
"step": 33500
},
{
"epoch": 0.53,
"learning_rate": 4.8345089299792914e-05,
"loss": 0.9303,
"step": 34000
},
{
"epoch": 0.54,
"learning_rate": 4.831880919593395e-05,
"loss": 0.9355,
"step": 34500
},
{
"epoch": 0.55,
"learning_rate": 4.829252909207497e-05,
"loss": 0.9463,
"step": 35000
},
{
"epoch": 0.56,
"learning_rate": 4.8266248988216006e-05,
"loss": 0.9333,
"step": 35500
},
{
"epoch": 0.57,
"learning_rate": 4.823996888435703e-05,
"loss": 0.9293,
"step": 36000
},
{
"epoch": 0.57,
"learning_rate": 4.8213688780498065e-05,
"loss": 0.9272,
"step": 36500
},
{
"epoch": 0.58,
"learning_rate": 4.818740867663909e-05,
"loss": 0.928,
"step": 37000
},
{
"epoch": 0.59,
"learning_rate": 4.8161128572780124e-05,
"loss": 0.9346,
"step": 37500
},
{
"epoch": 0.6,
"learning_rate": 4.813484846892115e-05,
"loss": 0.9361,
"step": 38000
},
{
"epoch": 0.61,
"learning_rate": 4.81086209252699e-05,
"loss": 0.9197,
"step": 38500
},
{
"epoch": 0.61,
"learning_rate": 4.8082393381618646e-05,
"loss": 0.9289,
"step": 39000
},
{
"epoch": 0.62,
"learning_rate": 4.805611327775967e-05,
"loss": 0.9232,
"step": 39500
},
{
"epoch": 0.63,
"learning_rate": 4.8029833173900705e-05,
"loss": 0.9134,
"step": 40000
},
{
"epoch": 0.64,
"learning_rate": 4.800355307004174e-05,
"loss": 0.9232,
"step": 40500
},
{
"epoch": 0.64,
"learning_rate": 4.7977272966182764e-05,
"loss": 0.9315,
"step": 41000
},
{
"epoch": 0.65,
"learning_rate": 4.7950992862323796e-05,
"loss": 0.9152,
"step": 41500
},
{
"epoch": 0.66,
"learning_rate": 4.792471275846482e-05,
"loss": 0.9179,
"step": 42000
},
{
"epoch": 0.67,
"learning_rate": 4.7898485214813574e-05,
"loss": 0.914,
"step": 42500
},
{
"epoch": 0.68,
"learning_rate": 4.787225767116232e-05,
"loss": 0.9078,
"step": 43000
},
{
"epoch": 0.68,
"learning_rate": 4.7845977567303345e-05,
"loss": 0.9129,
"step": 43500
},
{
"epoch": 0.69,
"learning_rate": 4.781969746344438e-05,
"loss": 0.899,
"step": 44000
},
{
"epoch": 0.7,
"learning_rate": 4.779341735958541e-05,
"loss": 0.9003,
"step": 44500
},
{
"epoch": 0.71,
"learning_rate": 4.7767137255726436e-05,
"loss": 0.8976,
"step": 45000
},
{
"epoch": 0.72,
"learning_rate": 4.774085715186747e-05,
"loss": 0.9007,
"step": 45500
},
{
"epoch": 0.72,
"learning_rate": 4.7714577048008495e-05,
"loss": 0.9053,
"step": 46000
},
{
"epoch": 0.73,
"learning_rate": 4.768829694414952e-05,
"loss": 0.892,
"step": 46500
},
{
"epoch": 0.74,
"learning_rate": 4.766201684029056e-05,
"loss": 0.9049,
"step": 47000
},
{
"epoch": 0.75,
"learning_rate": 4.763573673643159e-05,
"loss": 0.8988,
"step": 47500
},
{
"epoch": 0.75,
"learning_rate": 4.760956175298805e-05,
"loss": 0.9086,
"step": 48000
},
{
"epoch": 0.76,
"learning_rate": 4.758328164912908e-05,
"loss": 0.903,
"step": 48500
},
{
"epoch": 0.77,
"learning_rate": 4.755700154527011e-05,
"loss": 0.895,
"step": 49000
},
{
"epoch": 0.78,
"learning_rate": 4.753072144141114e-05,
"loss": 0.8982,
"step": 49500
},
{
"epoch": 0.79,
"learning_rate": 4.750444133755217e-05,
"loss": 0.9047,
"step": 50000
},
{
"epoch": 0.79,
"learning_rate": 4.7478161233693194e-05,
"loss": 0.9088,
"step": 50500
},
{
"epoch": 0.8,
"learning_rate": 4.745188112983423e-05,
"loss": 0.892,
"step": 51000
},
{
"epoch": 0.81,
"learning_rate": 4.742560102597526e-05,
"loss": 0.89,
"step": 51500
},
{
"epoch": 0.82,
"learning_rate": 4.7399320922116286e-05,
"loss": 0.8909,
"step": 52000
},
{
"epoch": 0.83,
"learning_rate": 4.737309337846503e-05,
"loss": 0.8792,
"step": 52500
},
{
"epoch": 0.83,
"learning_rate": 4.734681327460606e-05,
"loss": 0.8783,
"step": 53000
},
{
"epoch": 0.84,
"learning_rate": 4.7320533170747096e-05,
"loss": 0.8869,
"step": 53500
},
{
"epoch": 0.85,
"learning_rate": 4.729425306688812e-05,
"loss": 0.8873,
"step": 54000
},
{
"epoch": 0.86,
"learning_rate": 4.726797296302915e-05,
"loss": 0.8927,
"step": 54500
},
{
"epoch": 0.86,
"learning_rate": 4.72417454193779e-05,
"loss": 0.8817,
"step": 55000
},
{
"epoch": 0.87,
"learning_rate": 4.721546531551893e-05,
"loss": 0.8746,
"step": 55500
},
{
"epoch": 0.88,
"learning_rate": 4.718918521165996e-05,
"loss": 0.8845,
"step": 56000
},
{
"epoch": 0.89,
"learning_rate": 4.7162905107800984e-05,
"loss": 0.8768,
"step": 56500
},
{
"epoch": 0.9,
"learning_rate": 4.713662500394202e-05,
"loss": 0.8834,
"step": 57000
},
{
"epoch": 0.9,
"learning_rate": 4.711034490008305e-05,
"loss": 0.8769,
"step": 57500
},
{
"epoch": 0.91,
"learning_rate": 4.7084064796224076e-05,
"loss": 0.865,
"step": 58000
},
{
"epoch": 0.92,
"learning_rate": 4.705783725257282e-05,
"loss": 0.8786,
"step": 58500
},
{
"epoch": 0.93,
"learning_rate": 4.7031557148713854e-05,
"loss": 0.885,
"step": 59000
},
{
"epoch": 0.94,
"learning_rate": 4.7005277044854886e-05,
"loss": 0.8804,
"step": 59500
},
{
"epoch": 0.94,
"learning_rate": 4.697899694099591e-05,
"loss": 0.8834,
"step": 60000
},
{
"epoch": 0.95,
"learning_rate": 4.6952716837136945e-05,
"loss": 0.8698,
"step": 60500
},
{
"epoch": 0.96,
"learning_rate": 4.692643673327797e-05,
"loss": 0.8605,
"step": 61000
},
{
"epoch": 0.97,
"learning_rate": 4.6900156629419004e-05,
"loss": 0.8721,
"step": 61500
},
{
"epoch": 0.98,
"learning_rate": 4.687387652556003e-05,
"loss": 0.8914,
"step": 62000
},
{
"epoch": 0.98,
"learning_rate": 4.684759642170106e-05,
"loss": 0.8743,
"step": 62500
},
{
"epoch": 0.99,
"learning_rate": 4.682131631784209e-05,
"loss": 0.8677,
"step": 63000
},
{
"epoch": 1.0,
"learning_rate": 4.679514133439856e-05,
"loss": 0.8673,
"step": 63500
},
{
"epoch": 1.0,
"eval_bleu": 52.6771,
"eval_gen_len": 15.2057,
"eval_loss": 0.7307217121124268,
"eval_runtime": 9605.9098,
"eval_samples_per_second": 13.439,
"eval_steps_per_second": 1.68,
"step": 63586
},
{
"epoch": 1.01,
"learning_rate": 4.6768913790747304e-05,
"loss": 0.7651,
"step": 64000
},
{
"epoch": 1.01,
"learning_rate": 4.674263368688833e-05,
"loss": 0.7382,
"step": 64500
},
{
"epoch": 1.02,
"learning_rate": 4.671635358302936e-05,
"loss": 0.7193,
"step": 65000
},
{
"epoch": 1.03,
"learning_rate": 4.6690073479170396e-05,
"loss": 0.7344,
"step": 65500
},
{
"epoch": 1.04,
"learning_rate": 4.666384593551914e-05,
"loss": 0.7191,
"step": 66000
},
{
"epoch": 1.05,
"learning_rate": 4.6637565831660166e-05,
"loss": 0.7228,
"step": 66500
},
{
"epoch": 1.05,
"learning_rate": 4.66112857278012e-05,
"loss": 0.7285,
"step": 67000
},
{
"epoch": 1.06,
"learning_rate": 4.658500562394223e-05,
"loss": 0.7373,
"step": 67500
},
{
"epoch": 1.07,
"learning_rate": 4.655872552008326e-05,
"loss": 0.7209,
"step": 68000
},
{
"epoch": 1.08,
"learning_rate": 4.6532445416224284e-05,
"loss": 0.7345,
"step": 68500
},
{
"epoch": 1.09,
"learning_rate": 4.6506217872573036e-05,
"loss": 0.7348,
"step": 69000
},
{
"epoch": 1.09,
"learning_rate": 4.647993776871407e-05,
"loss": 0.7391,
"step": 69500
},
{
"epoch": 1.1,
"learning_rate": 4.6453657664855094e-05,
"loss": 0.7367,
"step": 70000
},
{
"epoch": 1.11,
"learning_rate": 4.642737756099612e-05,
"loss": 0.7282,
"step": 70500
},
{
"epoch": 1.12,
"learning_rate": 4.640109745713715e-05,
"loss": 0.7379,
"step": 71000
},
{
"epoch": 1.12,
"learning_rate": 4.6374817353278186e-05,
"loss": 0.7313,
"step": 71500
},
{
"epoch": 1.13,
"learning_rate": 4.634853724941921e-05,
"loss": 0.7399,
"step": 72000
},
{
"epoch": 1.14,
"learning_rate": 4.6322257145560245e-05,
"loss": 0.731,
"step": 72500
},
{
"epoch": 1.15,
"learning_rate": 4.629597704170127e-05,
"loss": 0.7435,
"step": 73000
},
{
"epoch": 1.16,
"learning_rate": 4.62696969378423e-05,
"loss": 0.7361,
"step": 73500
},
{
"epoch": 1.16,
"learning_rate": 4.624341683398333e-05,
"loss": 0.7315,
"step": 74000
},
{
"epoch": 1.17,
"learning_rate": 4.621713673012436e-05,
"loss": 0.7404,
"step": 74500
},
{
"epoch": 1.18,
"learning_rate": 4.619090918647311e-05,
"loss": 0.7399,
"step": 75000
},
{
"epoch": 1.19,
"learning_rate": 4.616468164282186e-05,
"loss": 0.7378,
"step": 75500
},
{
"epoch": 1.2,
"learning_rate": 4.6138401538962885e-05,
"loss": 0.7391,
"step": 76000
},
{
"epoch": 1.2,
"learning_rate": 4.611212143510391e-05,
"loss": 0.7499,
"step": 76500
},
{
"epoch": 1.21,
"learning_rate": 4.6085841331244944e-05,
"loss": 0.7537,
"step": 77000
},
{
"epoch": 1.22,
"learning_rate": 4.605956122738597e-05,
"loss": 0.7423,
"step": 77500
},
{
"epoch": 1.23,
"learning_rate": 4.6033281123527e-05,
"loss": 0.7453,
"step": 78000
},
{
"epoch": 1.23,
"learning_rate": 4.6007001019668035e-05,
"loss": 0.7439,
"step": 78500
},
{
"epoch": 1.24,
"learning_rate": 4.598072091580906e-05,
"loss": 0.7446,
"step": 79000
},
{
"epoch": 1.25,
"learning_rate": 4.595444081195009e-05,
"loss": 0.7412,
"step": 79500
},
{
"epoch": 1.26,
"learning_rate": 4.592821326829884e-05,
"loss": 0.7399,
"step": 80000
},
{
"epoch": 1.27,
"learning_rate": 4.590193316443987e-05,
"loss": 0.7302,
"step": 80500
},
{
"epoch": 1.27,
"learning_rate": 4.58756530605809e-05,
"loss": 0.7431,
"step": 81000
},
{
"epoch": 1.28,
"learning_rate": 4.5849372956721924e-05,
"loss": 0.7437,
"step": 81500
},
{
"epoch": 1.29,
"learning_rate": 4.5823145413070675e-05,
"loss": 0.7443,
"step": 82000
},
{
"epoch": 1.3,
"learning_rate": 4.579686530921171e-05,
"loss": 0.7496,
"step": 82500
},
{
"epoch": 1.31,
"learning_rate": 4.577063776556045e-05,
"loss": 0.7395,
"step": 83000
},
{
"epoch": 1.31,
"learning_rate": 4.574435766170148e-05,
"loss": 0.7307,
"step": 83500
},
{
"epoch": 1.32,
"learning_rate": 4.571807755784251e-05,
"loss": 0.7467,
"step": 84000
},
{
"epoch": 1.33,
"learning_rate": 4.5691797453983544e-05,
"loss": 0.7474,
"step": 84500
},
{
"epoch": 1.34,
"learning_rate": 4.566551735012457e-05,
"loss": 0.7352,
"step": 85000
},
{
"epoch": 1.34,
"learning_rate": 4.5639237246265596e-05,
"loss": 0.7377,
"step": 85500
},
{
"epoch": 1.35,
"learning_rate": 4.561295714240663e-05,
"loss": 0.7389,
"step": 86000
},
{
"epoch": 1.36,
"learning_rate": 4.558667703854766e-05,
"loss": 0.7372,
"step": 86500
},
{
"epoch": 1.37,
"learning_rate": 4.556039693468869e-05,
"loss": 0.7488,
"step": 87000
},
{
"epoch": 1.38,
"learning_rate": 4.553416939103743e-05,
"loss": 0.7303,
"step": 87500
},
{
"epoch": 1.38,
"learning_rate": 4.5507889287178466e-05,
"loss": 0.7243,
"step": 88000
},
{
"epoch": 1.39,
"learning_rate": 4.54816091833195e-05,
"loss": 0.7318,
"step": 88500
},
{
"epoch": 1.4,
"learning_rate": 4.5455329079460524e-05,
"loss": 0.7421,
"step": 89000
},
{
"epoch": 1.41,
"learning_rate": 4.542904897560155e-05,
"loss": 0.7441,
"step": 89500
},
{
"epoch": 1.42,
"learning_rate": 4.540276887174258e-05,
"loss": 0.7388,
"step": 90000
},
{
"epoch": 1.42,
"learning_rate": 4.537659388829905e-05,
"loss": 0.7381,
"step": 90500
},
{
"epoch": 1.43,
"learning_rate": 4.535031378444008e-05,
"loss": 0.7402,
"step": 91000
},
{
"epoch": 1.44,
"learning_rate": 4.5324033680581106e-05,
"loss": 0.7377,
"step": 91500
},
{
"epoch": 1.45,
"learning_rate": 4.529775357672214e-05,
"loss": 0.7474,
"step": 92000
},
{
"epoch": 1.45,
"learning_rate": 4.527152603307088e-05,
"loss": 0.7473,
"step": 92500
},
{
"epoch": 1.46,
"learning_rate": 4.5245245929211916e-05,
"loss": 0.7345,
"step": 93000
},
{
"epoch": 1.47,
"learning_rate": 4.521896582535294e-05,
"loss": 0.7366,
"step": 93500
},
{
"epoch": 1.48,
"learning_rate": 4.5192685721493975e-05,
"loss": 0.7518,
"step": 94000
},
{
"epoch": 1.49,
"learning_rate": 4.516640561763501e-05,
"loss": 0.7454,
"step": 94500
},
{
"epoch": 1.49,
"learning_rate": 4.514017807398375e-05,
"loss": 0.7432,
"step": 95000
},
{
"epoch": 1.5,
"learning_rate": 4.511389797012478e-05,
"loss": 0.7448,
"step": 95500
},
{
"epoch": 1.51,
"learning_rate": 4.508761786626581e-05,
"loss": 0.7415,
"step": 96000
},
{
"epoch": 1.52,
"learning_rate": 4.5061337762406844e-05,
"loss": 0.7358,
"step": 96500
},
{
"epoch": 1.53,
"learning_rate": 4.503505765854787e-05,
"loss": 0.7491,
"step": 97000
},
{
"epoch": 1.53,
"learning_rate": 4.5008777554688896e-05,
"loss": 0.7429,
"step": 97500
},
{
"epoch": 1.54,
"learning_rate": 4.498249745082993e-05,
"loss": 0.748,
"step": 98000
},
{
"epoch": 1.55,
"learning_rate": 4.4956217346970955e-05,
"loss": 0.7467,
"step": 98500
},
{
"epoch": 1.56,
"learning_rate": 4.492993724311199e-05,
"loss": 0.7459,
"step": 99000
},
{
"epoch": 1.56,
"learning_rate": 4.490365713925302e-05,
"loss": 0.7402,
"step": 99500
},
{
"epoch": 1.57,
"learning_rate": 4.4877377035394046e-05,
"loss": 0.7532,
"step": 100000
},
{
"epoch": 1.58,
"learning_rate": 4.485120205195052e-05,
"loss": 0.7464,
"step": 100500
},
{
"epoch": 1.59,
"learning_rate": 4.482492194809154e-05,
"loss": 0.7441,
"step": 101000
},
{
"epoch": 1.6,
"learning_rate": 4.479869440444029e-05,
"loss": 0.7445,
"step": 101500
},
{
"epoch": 1.6,
"learning_rate": 4.4772414300581314e-05,
"loss": 0.7316,
"step": 102000
},
{
"epoch": 1.61,
"learning_rate": 4.474613419672235e-05,
"loss": 0.7437,
"step": 102500
},
{
"epoch": 1.62,
"learning_rate": 4.471985409286338e-05,
"loss": 0.7456,
"step": 103000
},
{
"epoch": 1.63,
"learning_rate": 4.4693573989004405e-05,
"loss": 0.7508,
"step": 103500
},
{
"epoch": 1.64,
"learning_rate": 4.466729388514543e-05,
"loss": 0.7436,
"step": 104000
},
{
"epoch": 1.64,
"learning_rate": 4.4641013781286464e-05,
"loss": 0.7413,
"step": 104500
},
{
"epoch": 1.65,
"learning_rate": 4.46147336774275e-05,
"loss": 0.7305,
"step": 105000
},
{
"epoch": 1.66,
"learning_rate": 4.458845357356852e-05,
"loss": 0.7496,
"step": 105500
},
{
"epoch": 1.67,
"learning_rate": 4.4562173469709555e-05,
"loss": 0.7467,
"step": 106000
},
{
"epoch": 1.67,
"learning_rate": 4.453589336585058e-05,
"loss": 0.7404,
"step": 106500
},
{
"epoch": 1.68,
"learning_rate": 4.4509613261991614e-05,
"loss": 0.74,
"step": 107000
},
{
"epoch": 1.69,
"learning_rate": 4.448333315813265e-05,
"loss": 0.7432,
"step": 107500
},
{
"epoch": 1.7,
"learning_rate": 4.445705305427367e-05,
"loss": 0.7401,
"step": 108000
},
{
"epoch": 1.71,
"learning_rate": 4.44307729504147e-05,
"loss": 0.7382,
"step": 108500
},
{
"epoch": 1.71,
"learning_rate": 4.440449284655573e-05,
"loss": 0.7352,
"step": 109000
},
{
"epoch": 1.72,
"learning_rate": 4.4378212742696765e-05,
"loss": 0.7293,
"step": 109500
},
{
"epoch": 1.73,
"learning_rate": 4.435193263883779e-05,
"loss": 0.742,
"step": 110000
},
{
"epoch": 1.74,
"learning_rate": 4.432565253497882e-05,
"loss": 0.7426,
"step": 110500
},
{
"epoch": 1.75,
"learning_rate": 4.429942499132757e-05,
"loss": 0.7428,
"step": 111000
},
{
"epoch": 1.75,
"learning_rate": 4.42731448874686e-05,
"loss": 0.7391,
"step": 111500
},
{
"epoch": 1.76,
"learning_rate": 4.4246917343817346e-05,
"loss": 0.7449,
"step": 112000
},
{
"epoch": 1.77,
"learning_rate": 4.422063723995837e-05,
"loss": 0.7377,
"step": 112500
},
{
"epoch": 1.78,
"learning_rate": 4.4194357136099405e-05,
"loss": 0.7468,
"step": 113000
},
{
"epoch": 1.78,
"learning_rate": 4.416807703224044e-05,
"loss": 0.7398,
"step": 113500
},
{
"epoch": 1.79,
"learning_rate": 4.4141796928381463e-05,
"loss": 0.7371,
"step": 114000
},
{
"epoch": 1.8,
"learning_rate": 4.411556938473021e-05,
"loss": 0.7371,
"step": 114500
},
{
"epoch": 1.81,
"learning_rate": 4.4089289280871234e-05,
"loss": 0.7443,
"step": 115000
},
{
"epoch": 1.82,
"learning_rate": 4.4063009177012274e-05,
"loss": 0.7375,
"step": 115500
},
{
"epoch": 1.82,
"learning_rate": 4.40367290731533e-05,
"loss": 0.7339,
"step": 116000
},
{
"epoch": 1.83,
"learning_rate": 4.4010448969294326e-05,
"loss": 0.7267,
"step": 116500
},
{
"epoch": 1.84,
"learning_rate": 4.398416886543536e-05,
"loss": 0.7378,
"step": 117000
},
{
"epoch": 1.85,
"learning_rate": 4.3957888761576385e-05,
"loss": 0.7313,
"step": 117500
},
{
"epoch": 1.86,
"learning_rate": 4.393160865771742e-05,
"loss": 0.739,
"step": 118000
},
{
"epoch": 1.86,
"learning_rate": 4.390532855385845e-05,
"loss": 0.7314,
"step": 118500
},
{
"epoch": 1.87,
"learning_rate": 4.3879048449999476e-05,
"loss": 0.7328,
"step": 119000
},
{
"epoch": 1.88,
"learning_rate": 4.385282090634822e-05,
"loss": 0.741,
"step": 119500
},
{
"epoch": 1.89,
"learning_rate": 4.382659336269697e-05,
"loss": 0.7419,
"step": 120000
},
{
"epoch": 1.9,
"learning_rate": 4.3800313258838e-05,
"loss": 0.7353,
"step": 120500
},
{
"epoch": 1.9,
"learning_rate": 4.377403315497903e-05,
"loss": 0.7315,
"step": 121000
},
{
"epoch": 1.91,
"learning_rate": 4.374775305112006e-05,
"loss": 0.7457,
"step": 121500
},
{
"epoch": 1.92,
"learning_rate": 4.372147294726109e-05,
"loss": 0.7442,
"step": 122000
},
{
"epoch": 1.93,
"learning_rate": 4.369519284340212e-05,
"loss": 0.7421,
"step": 122500
},
{
"epoch": 1.93,
"learning_rate": 4.366891273954315e-05,
"loss": 0.7298,
"step": 123000
},
{
"epoch": 1.94,
"learning_rate": 4.3642632635684175e-05,
"loss": 0.7387,
"step": 123500
},
{
"epoch": 1.95,
"learning_rate": 4.361635253182521e-05,
"loss": 0.7322,
"step": 124000
},
{
"epoch": 1.96,
"learning_rate": 4.359012498817395e-05,
"loss": 0.7308,
"step": 124500
},
{
"epoch": 1.97,
"learning_rate": 4.3563844884314985e-05,
"loss": 0.7491,
"step": 125000
},
{
"epoch": 1.97,
"learning_rate": 4.353756478045601e-05,
"loss": 0.7292,
"step": 125500
},
{
"epoch": 1.98,
"learning_rate": 4.3511284676597044e-05,
"loss": 0.7307,
"step": 126000
},
{
"epoch": 1.99,
"learning_rate": 4.348500457273808e-05,
"loss": 0.7399,
"step": 126500
},
{
"epoch": 2.0,
"learning_rate": 4.345877702908682e-05,
"loss": 0.7184,
"step": 127000
},
{
"epoch": 2.0,
"eval_bleu": 55.1744,
"eval_gen_len": 15.3621,
"eval_loss": 0.6736727952957153,
"eval_runtime": 9675.9055,
"eval_samples_per_second": 13.342,
"eval_steps_per_second": 1.668,
"step": 127173
},
{
"epoch": 2.01,
"learning_rate": 4.343249692522785e-05,
"loss": 0.6388,
"step": 127500
},
{
"epoch": 2.01,
"learning_rate": 4.340621682136888e-05,
"loss": 0.5765,
"step": 128000
},
{
"epoch": 2.02,
"learning_rate": 4.3379936717509913e-05,
"loss": 0.5839,
"step": 128500
},
{
"epoch": 2.03,
"learning_rate": 4.335365661365094e-05,
"loss": 0.5815,
"step": 129000
},
{
"epoch": 2.04,
"learning_rate": 4.3327376509791965e-05,
"loss": 0.5716,
"step": 129500
},
{
"epoch": 2.04,
"learning_rate": 4.3301096405933e-05,
"loss": 0.5826,
"step": 130000
},
{
"epoch": 2.05,
"learning_rate": 4.327481630207403e-05,
"loss": 0.5783,
"step": 130500
},
{
"epoch": 2.06,
"learning_rate": 4.3248588758422776e-05,
"loss": 0.5879,
"step": 131000
},
{
"epoch": 2.07,
"learning_rate": 4.32223086545638e-05,
"loss": 0.5824,
"step": 131500
},
{
"epoch": 2.08,
"learning_rate": 4.3196028550704835e-05,
"loss": 0.583,
"step": 132000
},
{
"epoch": 2.08,
"learning_rate": 4.316974844684587e-05,
"loss": 0.5797,
"step": 132500
},
{
"epoch": 2.09,
"learning_rate": 4.3143468342986893e-05,
"loss": 0.5963,
"step": 133000
},
{
"epoch": 2.1,
"learning_rate": 4.311724079933564e-05,
"loss": 0.5841,
"step": 133500
},
{
"epoch": 2.11,
"learning_rate": 4.309096069547667e-05,
"loss": 0.5863,
"step": 134000
},
{
"epoch": 2.12,
"learning_rate": 4.3064680591617704e-05,
"loss": 0.5947,
"step": 134500
},
{
"epoch": 2.12,
"learning_rate": 4.303840048775873e-05,
"loss": 0.5977,
"step": 135000
},
{
"epoch": 2.13,
"learning_rate": 4.3012172944107475e-05,
"loss": 0.5848,
"step": 135500
},
{
"epoch": 2.14,
"learning_rate": 4.298589284024851e-05,
"loss": 0.5972,
"step": 136000
},
{
"epoch": 2.15,
"learning_rate": 4.295961273638954e-05,
"loss": 0.5979,
"step": 136500
},
{
"epoch": 2.15,
"learning_rate": 4.2933385192738285e-05,
"loss": 0.5879,
"step": 137000
},
{
"epoch": 2.16,
"learning_rate": 4.290710508887931e-05,
"loss": 0.6024,
"step": 137500
},
{
"epoch": 2.17,
"learning_rate": 4.288082498502034e-05,
"loss": 0.5893,
"step": 138000
},
{
"epoch": 2.18,
"learning_rate": 4.285459744136909e-05,
"loss": 0.6006,
"step": 138500
},
{
"epoch": 2.19,
"learning_rate": 4.282831733751012e-05,
"loss": 0.5983,
"step": 139000
},
{
"epoch": 2.19,
"learning_rate": 4.280203723365115e-05,
"loss": 0.6009,
"step": 139500
},
{
"epoch": 2.2,
"learning_rate": 4.2775757129792173e-05,
"loss": 0.6056,
"step": 140000
},
{
"epoch": 2.21,
"learning_rate": 4.274947702593321e-05,
"loss": 0.5942,
"step": 140500
},
{
"epoch": 2.22,
"learning_rate": 4.272330204248967e-05,
"loss": 0.5949,
"step": 141000
},
{
"epoch": 2.23,
"learning_rate": 4.26970219386307e-05,
"loss": 0.5972,
"step": 141500
},
{
"epoch": 2.23,
"learning_rate": 4.2670741834771735e-05,
"loss": 0.6073,
"step": 142000
},
{
"epoch": 2.24,
"learning_rate": 4.264446173091276e-05,
"loss": 0.6031,
"step": 142500
},
{
"epoch": 2.25,
"learning_rate": 4.2618181627053794e-05,
"loss": 0.5975,
"step": 143000
},
{
"epoch": 2.26,
"learning_rate": 4.259190152319482e-05,
"loss": 0.5965,
"step": 143500
},
{
"epoch": 2.26,
"learning_rate": 4.2565621419335846e-05,
"loss": 0.5991,
"step": 144000
},
{
"epoch": 2.27,
"learning_rate": 4.2539341315476886e-05,
"loss": 0.597,
"step": 144500
},
{
"epoch": 2.28,
"learning_rate": 4.251306121161791e-05,
"loss": 0.5909,
"step": 145000
},
{
"epoch": 2.29,
"learning_rate": 4.248678110775894e-05,
"loss": 0.6014,
"step": 145500
},
{
"epoch": 2.3,
"learning_rate": 4.246050100389997e-05,
"loss": 0.6016,
"step": 146000
},
{
"epoch": 2.3,
"learning_rate": 4.2434220900041e-05,
"loss": 0.6034,
"step": 146500
},
{
"epoch": 2.31,
"learning_rate": 4.240794079618203e-05,
"loss": 0.6033,
"step": 147000
},
{
"epoch": 2.32,
"learning_rate": 4.238166069232306e-05,
"loss": 0.6034,
"step": 147500
},
{
"epoch": 2.33,
"learning_rate": 4.235538058846409e-05,
"loss": 0.6123,
"step": 148000
},
{
"epoch": 2.34,
"learning_rate": 4.2329100484605114e-05,
"loss": 0.6087,
"step": 148500
},
{
"epoch": 2.34,
"learning_rate": 4.230282038074615e-05,
"loss": 0.6092,
"step": 149000
},
{
"epoch": 2.35,
"learning_rate": 4.227654027688718e-05,
"loss": 0.6045,
"step": 149500
},
{
"epoch": 2.36,
"learning_rate": 4.2250312733235925e-05,
"loss": 0.5975,
"step": 150000
},
{
"epoch": 2.37,
"learning_rate": 4.2224137749792395e-05,
"loss": 0.6023,
"step": 150500
},
{
"epoch": 2.37,
"learning_rate": 4.219785764593342e-05,
"loss": 0.6035,
"step": 151000
},
{
"epoch": 2.38,
"learning_rate": 4.2171630102282166e-05,
"loss": 0.5928,
"step": 151500
},
{
"epoch": 2.39,
"learning_rate": 4.214534999842319e-05,
"loss": 0.6116,
"step": 152000
},
{
"epoch": 2.4,
"learning_rate": 4.2119069894564225e-05,
"loss": 0.6186,
"step": 152500
},
{
"epoch": 2.41,
"learning_rate": 4.209278979070526e-05,
"loss": 0.6025,
"step": 153000
},
{
"epoch": 2.41,
"learning_rate": 4.2066509686846283e-05,
"loss": 0.5997,
"step": 153500
},
{
"epoch": 2.42,
"learning_rate": 4.204022958298731e-05,
"loss": 0.6094,
"step": 154000
},
{
"epoch": 2.43,
"learning_rate": 4.201394947912834e-05,
"loss": 0.6109,
"step": 154500
},
{
"epoch": 2.44,
"learning_rate": 4.1987669375269375e-05,
"loss": 0.5984,
"step": 155000
},
{
"epoch": 2.45,
"learning_rate": 4.19613892714104e-05,
"loss": 0.6177,
"step": 155500
},
{
"epoch": 2.45,
"learning_rate": 4.1935109167551434e-05,
"loss": 0.6136,
"step": 156000
},
{
"epoch": 2.46,
"learning_rate": 4.190882906369246e-05,
"loss": 0.6128,
"step": 156500
},
{
"epoch": 2.47,
"learning_rate": 4.188260152004121e-05,
"loss": 0.6172,
"step": 157000
},
{
"epoch": 2.48,
"learning_rate": 4.185632141618224e-05,
"loss": 0.6041,
"step": 157500
},
{
"epoch": 2.48,
"learning_rate": 4.183004131232327e-05,
"loss": 0.6154,
"step": 158000
},
{
"epoch": 2.49,
"learning_rate": 4.1803761208464296e-05,
"loss": 0.6094,
"step": 158500
},
{
"epoch": 2.5,
"learning_rate": 4.177748110460533e-05,
"loss": 0.6059,
"step": 159000
},
{
"epoch": 2.51,
"learning_rate": 4.175120100074636e-05,
"loss": 0.6089,
"step": 159500
},
{
"epoch": 2.52,
"learning_rate": 4.1724973457095107e-05,
"loss": 0.6122,
"step": 160000
},
{
"epoch": 2.52,
"learning_rate": 4.169874591344385e-05,
"loss": 0.6093,
"step": 160500
},
{
"epoch": 2.53,
"learning_rate": 4.1672465809584884e-05,
"loss": 0.6174,
"step": 161000
},
{
"epoch": 2.54,
"learning_rate": 4.164618570572591e-05,
"loss": 0.6301,
"step": 161500
},
{
"epoch": 2.55,
"learning_rate": 4.161990560186694e-05,
"loss": 0.6204,
"step": 162000
},
{
"epoch": 2.56,
"learning_rate": 4.159362549800797e-05,
"loss": 0.6207,
"step": 162500
},
{
"epoch": 2.56,
"learning_rate": 4.1567345394149e-05,
"loss": 0.6102,
"step": 163000
},
{
"epoch": 2.57,
"learning_rate": 4.154106529029003e-05,
"loss": 0.6088,
"step": 163500
},
{
"epoch": 2.58,
"learning_rate": 4.151478518643106e-05,
"loss": 0.6231,
"step": 164000
},
{
"epoch": 2.59,
"learning_rate": 4.1488505082572087e-05,
"loss": 0.6195,
"step": 164500
},
{
"epoch": 2.59,
"learning_rate": 4.146222497871311e-05,
"loss": 0.6214,
"step": 165000
},
{
"epoch": 2.6,
"learning_rate": 4.1435997435061864e-05,
"loss": 0.624,
"step": 165500
},
{
"epoch": 2.61,
"learning_rate": 4.14097173312029e-05,
"loss": 0.6218,
"step": 166000
},
{
"epoch": 2.62,
"learning_rate": 4.138343722734392e-05,
"loss": 0.6084,
"step": 166500
},
{
"epoch": 2.63,
"learning_rate": 4.135715712348495e-05,
"loss": 0.6155,
"step": 167000
},
{
"epoch": 2.63,
"learning_rate": 4.133087701962599e-05,
"loss": 0.6153,
"step": 167500
},
{
"epoch": 2.64,
"learning_rate": 4.1304596915767015e-05,
"loss": 0.6224,
"step": 168000
},
{
"epoch": 2.65,
"learning_rate": 4.127831681190804e-05,
"loss": 0.6132,
"step": 168500
},
{
"epoch": 2.66,
"learning_rate": 4.125203670804907e-05,
"loss": 0.6136,
"step": 169000
},
{
"epoch": 2.67,
"learning_rate": 4.12257566041901e-05,
"loss": 0.6097,
"step": 169500
},
{
"epoch": 2.67,
"learning_rate": 4.119952906053885e-05,
"loss": 0.6226,
"step": 170000
},
{
"epoch": 2.68,
"learning_rate": 4.117324895667988e-05,
"loss": 0.6152,
"step": 170500
},
{
"epoch": 2.69,
"learning_rate": 4.114696885282091e-05,
"loss": 0.6188,
"step": 171000
},
{
"epoch": 2.7,
"learning_rate": 4.112074130916966e-05,
"loss": 0.6199,
"step": 171500
},
{
"epoch": 2.7,
"learning_rate": 4.109446120531069e-05,
"loss": 0.626,
"step": 172000
},
{
"epoch": 2.71,
"learning_rate": 4.106823366165943e-05,
"loss": 0.6256,
"step": 172500
},
{
"epoch": 2.72,
"learning_rate": 4.104195355780046e-05,
"loss": 0.6169,
"step": 173000
},
{
"epoch": 2.73,
"learning_rate": 4.10156734539415e-05,
"loss": 0.6167,
"step": 173500
},
{
"epoch": 2.74,
"learning_rate": 4.0989393350082524e-05,
"loss": 0.6147,
"step": 174000
},
{
"epoch": 2.74,
"learning_rate": 4.096311324622355e-05,
"loss": 0.621,
"step": 174500
},
{
"epoch": 2.75,
"learning_rate": 4.0936833142364576e-05,
"loss": 0.6063,
"step": 175000
},
{
"epoch": 2.76,
"learning_rate": 4.091055303850561e-05,
"loss": 0.6216,
"step": 175500
},
{
"epoch": 2.77,
"learning_rate": 4.088427293464664e-05,
"loss": 0.6307,
"step": 176000
},
{
"epoch": 2.78,
"learning_rate": 4.085799283078767e-05,
"loss": 0.6313,
"step": 176500
},
{
"epoch": 2.78,
"learning_rate": 4.08317127269287e-05,
"loss": 0.6286,
"step": 177000
},
{
"epoch": 2.79,
"learning_rate": 4.0805432623069726e-05,
"loss": 0.6109,
"step": 177500
},
{
"epoch": 2.8,
"learning_rate": 4.077915251921076e-05,
"loss": 0.6175,
"step": 178000
},
{
"epoch": 2.81,
"learning_rate": 4.075287241535179e-05,
"loss": 0.6211,
"step": 178500
},
{
"epoch": 2.82,
"learning_rate": 4.072659231149282e-05,
"loss": 0.6227,
"step": 179000
},
{
"epoch": 2.82,
"learning_rate": 4.0700312207633844e-05,
"loss": 0.6254,
"step": 179500
},
{
"epoch": 2.83,
"learning_rate": 4.0674032103774877e-05,
"loss": 0.6212,
"step": 180000
},
{
"epoch": 2.84,
"learning_rate": 4.064775199991591e-05,
"loss": 0.6257,
"step": 180500
},
{
"epoch": 2.85,
"learning_rate": 4.0621471896056935e-05,
"loss": 0.6255,
"step": 181000
},
{
"epoch": 2.85,
"learning_rate": 4.059519179219797e-05,
"loss": 0.622,
"step": 181500
},
{
"epoch": 2.86,
"learning_rate": 4.0568911688338994e-05,
"loss": 0.6278,
"step": 182000
},
{
"epoch": 2.87,
"learning_rate": 4.0542684144687746e-05,
"loss": 0.6298,
"step": 182500
},
{
"epoch": 2.88,
"learning_rate": 4.051640404082877e-05,
"loss": 0.6197,
"step": 183000
},
{
"epoch": 2.89,
"learning_rate": 4.0490176497177517e-05,
"loss": 0.6169,
"step": 183500
},
{
"epoch": 2.89,
"learning_rate": 4.046389639331855e-05,
"loss": 0.6148,
"step": 184000
},
{
"epoch": 2.9,
"learning_rate": 4.043761628945958e-05,
"loss": 0.6153,
"step": 184500
},
{
"epoch": 2.91,
"learning_rate": 4.041133618560061e-05,
"loss": 0.622,
"step": 185000
},
{
"epoch": 2.92,
"learning_rate": 4.0385056081741634e-05,
"loss": 0.6291,
"step": 185500
},
{
"epoch": 2.93,
"learning_rate": 4.035882853809038e-05,
"loss": 0.6172,
"step": 186000
},
{
"epoch": 2.93,
"learning_rate": 4.033254843423142e-05,
"loss": 0.6156,
"step": 186500
},
{
"epoch": 2.94,
"learning_rate": 4.0306268330372445e-05,
"loss": 0.6274,
"step": 187000
},
{
"epoch": 2.95,
"learning_rate": 4.028004078672119e-05,
"loss": 0.6319,
"step": 187500
},
{
"epoch": 2.96,
"learning_rate": 4.025381324306994e-05,
"loss": 0.6218,
"step": 188000
},
{
"epoch": 2.96,
"learning_rate": 4.022753313921097e-05,
"loss": 0.6184,
"step": 188500
},
{
"epoch": 2.97,
"learning_rate": 4.0201253035352e-05,
"loss": 0.6275,
"step": 189000
},
{
"epoch": 2.98,
"learning_rate": 4.0174972931493026e-05,
"loss": 0.6218,
"step": 189500
},
{
"epoch": 2.99,
"learning_rate": 4.014869282763405e-05,
"loss": 0.6179,
"step": 190000
},
{
"epoch": 3.0,
"learning_rate": 4.01224652839828e-05,
"loss": 0.6294,
"step": 190500
},
{
"epoch": 3.0,
"eval_bleu": 55.7536,
"eval_gen_len": 15.2192,
"eval_loss": 0.6636257171630859,
"eval_runtime": 9606.0302,
"eval_samples_per_second": 13.439,
"eval_steps_per_second": 1.68,
"step": 190759
},
{
"epoch": 3.0,
"learning_rate": 4.0096185180123836e-05,
"loss": 0.551,
"step": 191000
},
{
"epoch": 3.01,
"learning_rate": 4.006990507626486e-05,
"loss": 0.4652,
"step": 191500
},
{
"epoch": 3.02,
"learning_rate": 4.004362497240589e-05,
"loss": 0.4579,
"step": 192000
},
{
"epoch": 3.03,
"learning_rate": 4.001734486854692e-05,
"loss": 0.4586,
"step": 192500
},
{
"epoch": 3.04,
"learning_rate": 3.9991064764687954e-05,
"loss": 0.4651,
"step": 193000
},
{
"epoch": 3.04,
"learning_rate": 3.99648372210367e-05,
"loss": 0.4688,
"step": 193500
},
{
"epoch": 3.05,
"learning_rate": 3.9938557117177725e-05,
"loss": 0.4685,
"step": 194000
},
{
"epoch": 3.06,
"learning_rate": 3.991227701331876e-05,
"loss": 0.4687,
"step": 194500
},
{
"epoch": 3.07,
"learning_rate": 3.988599690945979e-05,
"loss": 0.4672,
"step": 195000
},
{
"epoch": 3.07,
"learning_rate": 3.9859716805600816e-05,
"loss": 0.4693,
"step": 195500
},
{
"epoch": 3.08,
"learning_rate": 3.983348926194956e-05,
"loss": 0.4741,
"step": 196000
},
{
"epoch": 3.09,
"learning_rate": 3.9807209158090594e-05,
"loss": 0.4708,
"step": 196500
},
{
"epoch": 3.1,
"learning_rate": 3.9780929054231627e-05,
"loss": 0.4696,
"step": 197000
},
{
"epoch": 3.11,
"learning_rate": 3.975464895037265e-05,
"loss": 0.469,
"step": 197500
},
{
"epoch": 3.11,
"learning_rate": 3.9728368846513685e-05,
"loss": 0.4665,
"step": 198000
},
{
"epoch": 3.12,
"learning_rate": 3.970208874265471e-05,
"loss": 0.4742,
"step": 198500
},
{
"epoch": 3.13,
"learning_rate": 3.9675808638795744e-05,
"loss": 0.4854,
"step": 199000
},
{
"epoch": 3.14,
"learning_rate": 3.964952853493677e-05,
"loss": 0.4816,
"step": 199500
},
{
"epoch": 3.15,
"learning_rate": 3.96232484310778e-05,
"loss": 0.4851,
"step": 200000
},
{
"epoch": 3.15,
"learning_rate": 3.959702088742655e-05,
"loss": 0.4814,
"step": 200500
},
{
"epoch": 3.16,
"learning_rate": 3.957074078356758e-05,
"loss": 0.4766,
"step": 201000
},
{
"epoch": 3.17,
"learning_rate": 3.9544513239916325e-05,
"loss": 0.4869,
"step": 201500
},
{
"epoch": 3.18,
"learning_rate": 3.951823313605735e-05,
"loss": 0.4754,
"step": 202000
},
{
"epoch": 3.18,
"learning_rate": 3.9491953032198384e-05,
"loss": 0.4805,
"step": 202500
},
{
"epoch": 3.19,
"learning_rate": 3.946567292833942e-05,
"loss": 0.4876,
"step": 203000
},
{
"epoch": 3.2,
"learning_rate": 3.943939282448044e-05,
"loss": 0.4936,
"step": 203500
},
{
"epoch": 3.21,
"learning_rate": 3.9413112720621476e-05,
"loss": 0.4869,
"step": 204000
},
{
"epoch": 3.22,
"learning_rate": 3.93868326167625e-05,
"loss": 0.4943,
"step": 204500
},
{
"epoch": 3.22,
"learning_rate": 3.9360552512903535e-05,
"loss": 0.4947,
"step": 205000
},
{
"epoch": 3.23,
"learning_rate": 3.933427240904457e-05,
"loss": 0.4947,
"step": 205500
},
{
"epoch": 3.24,
"learning_rate": 3.930799230518559e-05,
"loss": 0.4849,
"step": 206000
},
{
"epoch": 3.25,
"learning_rate": 3.928171220132662e-05,
"loss": 0.4994,
"step": 206500
},
{
"epoch": 3.26,
"learning_rate": 3.925548465767537e-05,
"loss": 0.4892,
"step": 207000
},
{
"epoch": 3.26,
"learning_rate": 3.9229204553816404e-05,
"loss": 0.4856,
"step": 207500
},
{
"epoch": 3.27,
"learning_rate": 3.920292444995743e-05,
"loss": 0.4945,
"step": 208000
},
{
"epoch": 3.28,
"learning_rate": 3.9176696906306175e-05,
"loss": 0.4906,
"step": 208500
},
{
"epoch": 3.29,
"learning_rate": 3.9150469362654926e-05,
"loss": 0.4948,
"step": 209000
},
{
"epoch": 3.29,
"learning_rate": 3.912418925879595e-05,
"loss": 0.486,
"step": 209500
},
{
"epoch": 3.3,
"learning_rate": 3.9097909154936985e-05,
"loss": 0.4957,
"step": 210000
},
{
"epoch": 3.31,
"learning_rate": 3.907162905107801e-05,
"loss": 0.4912,
"step": 210500
},
{
"epoch": 3.32,
"learning_rate": 3.904540150742676e-05,
"loss": 0.4905,
"step": 211000
},
{
"epoch": 3.33,
"learning_rate": 3.901912140356779e-05,
"loss": 0.5008,
"step": 211500
},
{
"epoch": 3.33,
"learning_rate": 3.899284129970882e-05,
"loss": 0.4983,
"step": 212000
},
{
"epoch": 3.34,
"learning_rate": 3.896656119584985e-05,
"loss": 0.4988,
"step": 212500
},
{
"epoch": 3.35,
"learning_rate": 3.894028109199087e-05,
"loss": 0.4952,
"step": 213000
},
{
"epoch": 3.36,
"learning_rate": 3.8914000988131906e-05,
"loss": 0.4925,
"step": 213500
},
{
"epoch": 3.37,
"learning_rate": 3.888772088427294e-05,
"loss": 0.4992,
"step": 214000
},
{
"epoch": 3.37,
"learning_rate": 3.8861440780413965e-05,
"loss": 0.5013,
"step": 214500
},
{
"epoch": 3.38,
"learning_rate": 3.883521323676271e-05,
"loss": 0.5053,
"step": 215000
},
{
"epoch": 3.39,
"learning_rate": 3.880893313290374e-05,
"loss": 0.5039,
"step": 215500
},
{
"epoch": 3.4,
"learning_rate": 3.8782653029044775e-05,
"loss": 0.4962,
"step": 216000
},
{
"epoch": 3.4,
"learning_rate": 3.87563729251858e-05,
"loss": 0.4977,
"step": 216500
},
{
"epoch": 3.41,
"learning_rate": 3.873009282132683e-05,
"loss": 0.495,
"step": 217000
},
{
"epoch": 3.42,
"learning_rate": 3.870381271746786e-05,
"loss": 0.5101,
"step": 217500
},
{
"epoch": 3.43,
"learning_rate": 3.867753261360889e-05,
"loss": 0.4998,
"step": 218000
},
{
"epoch": 3.44,
"learning_rate": 3.865125250974992e-05,
"loss": 0.5069,
"step": 218500
},
{
"epoch": 3.44,
"learning_rate": 3.862497240589095e-05,
"loss": 0.5002,
"step": 219000
},
{
"epoch": 3.45,
"learning_rate": 3.859869230203198e-05,
"loss": 0.5035,
"step": 219500
},
{
"epoch": 3.46,
"learning_rate": 3.857241219817301e-05,
"loss": 0.5015,
"step": 220000
},
{
"epoch": 3.47,
"learning_rate": 3.8546132094314036e-05,
"loss": 0.5082,
"step": 220500
},
{
"epoch": 3.48,
"learning_rate": 3.851985199045507e-05,
"loss": 0.5069,
"step": 221000
},
{
"epoch": 3.48,
"learning_rate": 3.8493571886596095e-05,
"loss": 0.4975,
"step": 221500
},
{
"epoch": 3.49,
"learning_rate": 3.846729178273713e-05,
"loss": 0.512,
"step": 222000
},
{
"epoch": 3.5,
"learning_rate": 3.844111679929359e-05,
"loss": 0.5093,
"step": 222500
},
{
"epoch": 3.51,
"learning_rate": 3.8414889255642337e-05,
"loss": 0.5142,
"step": 223000
},
{
"epoch": 3.51,
"learning_rate": 3.838860915178337e-05,
"loss": 0.5048,
"step": 223500
},
{
"epoch": 3.52,
"learning_rate": 3.83623290479244e-05,
"loss": 0.5055,
"step": 224000
},
{
"epoch": 3.53,
"learning_rate": 3.833604894406543e-05,
"loss": 0.5102,
"step": 224500
},
{
"epoch": 3.54,
"learning_rate": 3.8309768840206454e-05,
"loss": 0.5213,
"step": 225000
},
{
"epoch": 3.55,
"learning_rate": 3.8283541296555206e-05,
"loss": 0.5063,
"step": 225500
},
{
"epoch": 3.55,
"learning_rate": 3.825726119269624e-05,
"loss": 0.5071,
"step": 226000
},
{
"epoch": 3.56,
"learning_rate": 3.823103364904498e-05,
"loss": 0.5042,
"step": 226500
},
{
"epoch": 3.57,
"learning_rate": 3.820475354518601e-05,
"loss": 0.5094,
"step": 227000
},
{
"epoch": 3.58,
"learning_rate": 3.817847344132704e-05,
"loss": 0.5141,
"step": 227500
},
{
"epoch": 3.59,
"learning_rate": 3.8152193337468075e-05,
"loss": 0.5031,
"step": 228000
},
{
"epoch": 3.59,
"learning_rate": 3.81259132336091e-05,
"loss": 0.5116,
"step": 228500
},
{
"epoch": 3.6,
"learning_rate": 3.809963312975013e-05,
"loss": 0.5178,
"step": 229000
},
{
"epoch": 3.61,
"learning_rate": 3.807335302589116e-05,
"loss": 0.5109,
"step": 229500
},
{
"epoch": 3.62,
"learning_rate": 3.804707292203219e-05,
"loss": 0.5091,
"step": 230000
},
{
"epoch": 3.62,
"learning_rate": 3.802084537838094e-05,
"loss": 0.507,
"step": 230500
},
{
"epoch": 3.63,
"learning_rate": 3.799456527452196e-05,
"loss": 0.508,
"step": 231000
},
{
"epoch": 3.64,
"learning_rate": 3.7968285170662996e-05,
"loss": 0.515,
"step": 231500
},
{
"epoch": 3.65,
"learning_rate": 3.794200506680403e-05,
"loss": 0.5169,
"step": 232000
},
{
"epoch": 3.66,
"learning_rate": 3.7915724962945055e-05,
"loss": 0.5249,
"step": 232500
},
{
"epoch": 3.66,
"learning_rate": 3.788944485908609e-05,
"loss": 0.5171,
"step": 233000
},
{
"epoch": 3.67,
"learning_rate": 3.7863164755227114e-05,
"loss": 0.5122,
"step": 233500
},
{
"epoch": 3.68,
"learning_rate": 3.783688465136814e-05,
"loss": 0.5101,
"step": 234000
},
{
"epoch": 3.69,
"learning_rate": 3.781060454750917e-05,
"loss": 0.5136,
"step": 234500
},
{
"epoch": 3.7,
"learning_rate": 3.7784324443650205e-05,
"loss": 0.5287,
"step": 235000
},
{
"epoch": 3.7,
"learning_rate": 3.775804433979123e-05,
"loss": 0.5192,
"step": 235500
},
{
"epoch": 3.71,
"learning_rate": 3.773176423593226e-05,
"loss": 0.5051,
"step": 236000
},
{
"epoch": 3.72,
"learning_rate": 3.770548413207329e-05,
"loss": 0.5229,
"step": 236500
},
{
"epoch": 3.73,
"learning_rate": 3.767920402821432e-05,
"loss": 0.5212,
"step": 237000
},
{
"epoch": 3.74,
"learning_rate": 3.765292392435535e-05,
"loss": 0.5236,
"step": 237500
},
{
"epoch": 3.74,
"learning_rate": 3.7626696380704094e-05,
"loss": 0.5146,
"step": 238000
},
{
"epoch": 3.75,
"learning_rate": 3.7600416276845126e-05,
"loss": 0.5245,
"step": 238500
},
{
"epoch": 3.76,
"learning_rate": 3.757418873319388e-05,
"loss": 0.5239,
"step": 239000
},
{
"epoch": 3.77,
"learning_rate": 3.7547908629334904e-05,
"loss": 0.5203,
"step": 239500
},
{
"epoch": 3.77,
"learning_rate": 3.752162852547593e-05,
"loss": 0.5051,
"step": 240000
},
{
"epoch": 3.78,
"learning_rate": 3.749534842161696e-05,
"loss": 0.519,
"step": 240500
},
{
"epoch": 3.79,
"learning_rate": 3.7469120877965714e-05,
"loss": 0.5311,
"step": 241000
},
{
"epoch": 3.8,
"learning_rate": 3.744284077410674e-05,
"loss": 0.5146,
"step": 241500
},
{
"epoch": 3.81,
"learning_rate": 3.7416560670247766e-05,
"loss": 0.523,
"step": 242000
},
{
"epoch": 3.81,
"learning_rate": 3.739033312659652e-05,
"loss": 0.5274,
"step": 242500
},
{
"epoch": 3.82,
"learning_rate": 3.736405302273755e-05,
"loss": 0.5089,
"step": 243000
},
{
"epoch": 3.83,
"learning_rate": 3.733777291887858e-05,
"loss": 0.5125,
"step": 243500
},
{
"epoch": 3.84,
"learning_rate": 3.73114928150196e-05,
"loss": 0.526,
"step": 244000
},
{
"epoch": 3.85,
"learning_rate": 3.7285212711160636e-05,
"loss": 0.5223,
"step": 244500
},
{
"epoch": 3.85,
"learning_rate": 3.725893260730167e-05,
"loss": 0.5305,
"step": 245000
},
{
"epoch": 3.86,
"learning_rate": 3.7232652503442694e-05,
"loss": 0.5268,
"step": 245500
},
{
"epoch": 3.87,
"learning_rate": 3.720637239958373e-05,
"loss": 0.5213,
"step": 246000
},
{
"epoch": 3.88,
"learning_rate": 3.718009229572475e-05,
"loss": 0.5203,
"step": 246500
},
{
"epoch": 3.88,
"learning_rate": 3.7153812191865786e-05,
"loss": 0.5339,
"step": 247000
},
{
"epoch": 3.89,
"learning_rate": 3.712753208800681e-05,
"loss": 0.5215,
"step": 247500
},
{
"epoch": 3.9,
"learning_rate": 3.7101251984147845e-05,
"loss": 0.5203,
"step": 248000
},
{
"epoch": 3.91,
"learning_rate": 3.707497188028887e-05,
"loss": 0.5162,
"step": 248500
},
{
"epoch": 3.92,
"learning_rate": 3.70486917764299e-05,
"loss": 0.5214,
"step": 249000
},
{
"epoch": 3.92,
"learning_rate": 3.7022411672570936e-05,
"loss": 0.5278,
"step": 249500
},
{
"epoch": 3.93,
"learning_rate": 3.699613156871196e-05,
"loss": 0.5251,
"step": 250000
},
{
"epoch": 3.94,
"learning_rate": 3.696985146485299e-05,
"loss": 0.5265,
"step": 250500
},
{
"epoch": 3.95,
"learning_rate": 3.694357136099402e-05,
"loss": 0.5222,
"step": 251000
},
{
"epoch": 3.96,
"learning_rate": 3.691729125713505e-05,
"loss": 0.518,
"step": 251500
},
{
"epoch": 3.96,
"learning_rate": 3.68910637134838e-05,
"loss": 0.5299,
"step": 252000
},
{
"epoch": 3.97,
"learning_rate": 3.6864783609624825e-05,
"loss": 0.5255,
"step": 252500
},
{
"epoch": 3.98,
"learning_rate": 3.683855606597357e-05,
"loss": 0.513,
"step": 253000
},
{
"epoch": 3.99,
"learning_rate": 3.681232852232232e-05,
"loss": 0.5178,
"step": 253500
},
{
"epoch": 3.99,
"learning_rate": 3.6786048418463354e-05,
"loss": 0.5313,
"step": 254000
},
{
"epoch": 4.0,
"eval_bleu": 55.9422,
"eval_gen_len": 15.2674,
"eval_loss": 0.6860370635986328,
"eval_runtime": 9607.1426,
"eval_samples_per_second": 13.437,
"eval_steps_per_second": 1.68,
"step": 254346
},
{
"epoch": 4.0,
"learning_rate": 3.675976831460438e-05,
"loss": 0.4774,
"step": 254500
},
{
"epoch": 4.01,
"learning_rate": 3.673354077095313e-05,
"loss": 0.3734,
"step": 255000
},
{
"epoch": 4.02,
"learning_rate": 3.6707313227301876e-05,
"loss": 0.3715,
"step": 255500
},
{
"epoch": 4.03,
"learning_rate": 3.66810331234429e-05,
"loss": 0.3741,
"step": 256000
},
{
"epoch": 4.03,
"learning_rate": 3.6654753019583935e-05,
"loss": 0.3715,
"step": 256500
},
{
"epoch": 4.04,
"learning_rate": 3.662847291572497e-05,
"loss": 0.3737,
"step": 257000
},
{
"epoch": 4.05,
"learning_rate": 3.6602192811865994e-05,
"loss": 0.3778,
"step": 257500
},
{
"epoch": 4.06,
"learning_rate": 3.657591270800703e-05,
"loss": 0.3739,
"step": 258000
},
{
"epoch": 4.07,
"learning_rate": 3.654963260414805e-05,
"loss": 0.3824,
"step": 258500
},
{
"epoch": 4.07,
"learning_rate": 3.652335250028908e-05,
"loss": 0.3834,
"step": 259000
},
{
"epoch": 4.08,
"learning_rate": 3.649707239643011e-05,
"loss": 0.3714,
"step": 259500
},
{
"epoch": 4.09,
"learning_rate": 3.6470792292571144e-05,
"loss": 0.3817,
"step": 260000
},
{
"epoch": 4.1,
"learning_rate": 3.644451218871217e-05,
"loss": 0.3781,
"step": 260500
},
{
"epoch": 4.1,
"learning_rate": 3.6418284645060915e-05,
"loss": 0.3836,
"step": 261000
},
{
"epoch": 4.11,
"learning_rate": 3.639200454120195e-05,
"loss": 0.3865,
"step": 261500
},
{
"epoch": 4.12,
"learning_rate": 3.636572443734298e-05,
"loss": 0.3837,
"step": 262000
},
{
"epoch": 4.13,
"learning_rate": 3.633944433348401e-05,
"loss": 0.384,
"step": 262500
},
{
"epoch": 4.14,
"learning_rate": 3.631316422962503e-05,
"loss": 0.3911,
"step": 263000
},
{
"epoch": 4.14,
"learning_rate": 3.6286884125766066e-05,
"loss": 0.3785,
"step": 263500
},
{
"epoch": 4.15,
"learning_rate": 3.62606040219071e-05,
"loss": 0.3826,
"step": 264000
},
{
"epoch": 4.16,
"learning_rate": 3.6234323918048124e-05,
"loss": 0.3827,
"step": 264500
},
{
"epoch": 4.17,
"learning_rate": 3.620809637439687e-05,
"loss": 0.393,
"step": 265000
},
{
"epoch": 4.18,
"learning_rate": 3.61818162705379e-05,
"loss": 0.3783,
"step": 265500
},
{
"epoch": 4.18,
"learning_rate": 3.6155536166678935e-05,
"loss": 0.3854,
"step": 266000
},
{
"epoch": 4.19,
"learning_rate": 3.612925606281996e-05,
"loss": 0.3824,
"step": 266500
},
{
"epoch": 4.2,
"learning_rate": 3.6103028519168706e-05,
"loss": 0.3858,
"step": 267000
},
{
"epoch": 4.21,
"learning_rate": 3.607674841530974e-05,
"loss": 0.3866,
"step": 267500
},
{
"epoch": 4.21,
"learning_rate": 3.605046831145077e-05,
"loss": 0.3862,
"step": 268000
},
{
"epoch": 4.22,
"learning_rate": 3.60241882075918e-05,
"loss": 0.3897,
"step": 268500
},
{
"epoch": 4.23,
"learning_rate": 3.599790810373283e-05,
"loss": 0.3894,
"step": 269000
},
{
"epoch": 4.24,
"learning_rate": 3.5971627999873856e-05,
"loss": 0.3895,
"step": 269500
},
{
"epoch": 4.25,
"learning_rate": 3.594534789601489e-05,
"loss": 0.3927,
"step": 270000
},
{
"epoch": 4.25,
"learning_rate": 3.5919067792155915e-05,
"loss": 0.4029,
"step": 270500
},
{
"epoch": 4.26,
"learning_rate": 3.589278768829695e-05,
"loss": 0.397,
"step": 271000
},
{
"epoch": 4.27,
"learning_rate": 3.5866507584437974e-05,
"loss": 0.3946,
"step": 271500
},
{
"epoch": 4.28,
"learning_rate": 3.5840227480579e-05,
"loss": 0.393,
"step": 272000
},
{
"epoch": 4.29,
"learning_rate": 3.581399993692775e-05,
"loss": 0.4006,
"step": 272500
},
{
"epoch": 4.29,
"learning_rate": 3.5787719833068784e-05,
"loss": 0.3956,
"step": 273000
},
{
"epoch": 4.3,
"learning_rate": 3.576143972920981e-05,
"loss": 0.3998,
"step": 273500
},
{
"epoch": 4.31,
"learning_rate": 3.5735159625350836e-05,
"loss": 0.3982,
"step": 274000
},
{
"epoch": 4.32,
"learning_rate": 3.5708879521491876e-05,
"loss": 0.3974,
"step": 274500
},
{
"epoch": 4.32,
"learning_rate": 3.56825994176329e-05,
"loss": 0.4028,
"step": 275000
},
{
"epoch": 4.33,
"learning_rate": 3.5656371873981646e-05,
"loss": 0.4118,
"step": 275500
},
{
"epoch": 4.34,
"learning_rate": 3.563009177012267e-05,
"loss": 0.4067,
"step": 276000
},
{
"epoch": 4.35,
"learning_rate": 3.560381166626371e-05,
"loss": 0.4008,
"step": 276500
},
{
"epoch": 4.36,
"learning_rate": 3.557753156240474e-05,
"loss": 0.4045,
"step": 277000
},
{
"epoch": 4.36,
"learning_rate": 3.5551251458545764e-05,
"loss": 0.4092,
"step": 277500
},
{
"epoch": 4.37,
"learning_rate": 3.55249713546868e-05,
"loss": 0.4109,
"step": 278000
},
{
"epoch": 4.38,
"learning_rate": 3.549869125082782e-05,
"loss": 0.409,
"step": 278500
},
{
"epoch": 4.39,
"learning_rate": 3.5472411146968856e-05,
"loss": 0.4024,
"step": 279000
},
{
"epoch": 4.4,
"learning_rate": 3.544613104310988e-05,
"loss": 0.4075,
"step": 279500
},
{
"epoch": 4.4,
"learning_rate": 3.5419850939250914e-05,
"loss": 0.4087,
"step": 280000
},
{
"epoch": 4.41,
"learning_rate": 3.539357083539194e-05,
"loss": 0.4039,
"step": 280500
},
{
"epoch": 4.42,
"learning_rate": 3.536729073153297e-05,
"loss": 0.4098,
"step": 281000
},
{
"epoch": 4.43,
"learning_rate": 3.5341010627674006e-05,
"loss": 0.4019,
"step": 281500
},
{
"epoch": 4.43,
"learning_rate": 3.531473052381503e-05,
"loss": 0.406,
"step": 282000
},
{
"epoch": 4.44,
"learning_rate": 3.528845041995606e-05,
"loss": 0.4078,
"step": 282500
},
{
"epoch": 4.45,
"learning_rate": 3.526217031609709e-05,
"loss": 0.409,
"step": 283000
},
{
"epoch": 4.46,
"learning_rate": 3.523594277244584e-05,
"loss": 0.4122,
"step": 283500
},
{
"epoch": 4.47,
"learning_rate": 3.520966266858687e-05,
"loss": 0.405,
"step": 284000
},
{
"epoch": 4.47,
"learning_rate": 3.5183382564727894e-05,
"loss": 0.4101,
"step": 284500
},
{
"epoch": 4.48,
"learning_rate": 3.515710246086893e-05,
"loss": 0.4218,
"step": 285000
},
{
"epoch": 4.49,
"learning_rate": 3.513082235700996e-05,
"loss": 0.4186,
"step": 285500
},
{
"epoch": 4.5,
"learning_rate": 3.5104542253150986e-05,
"loss": 0.4088,
"step": 286000
},
{
"epoch": 4.51,
"learning_rate": 3.507826214929202e-05,
"loss": 0.416,
"step": 286500
},
{
"epoch": 4.51,
"learning_rate": 3.5051982045433045e-05,
"loss": 0.4186,
"step": 287000
},
{
"epoch": 4.52,
"learning_rate": 3.502570194157407e-05,
"loss": 0.4146,
"step": 287500
},
{
"epoch": 4.53,
"learning_rate": 3.499947439792282e-05,
"loss": 0.4118,
"step": 288000
},
{
"epoch": 4.54,
"learning_rate": 3.497319429406385e-05,
"loss": 0.4141,
"step": 288500
},
{
"epoch": 4.54,
"learning_rate": 3.494691419020488e-05,
"loss": 0.4115,
"step": 289000
},
{
"epoch": 4.55,
"learning_rate": 3.492068664655363e-05,
"loss": 0.4136,
"step": 289500
},
{
"epoch": 4.56,
"learning_rate": 3.489440654269466e-05,
"loss": 0.4198,
"step": 290000
},
{
"epoch": 4.57,
"learning_rate": 3.4868126438835685e-05,
"loss": 0.4176,
"step": 290500
},
{
"epoch": 4.58,
"learning_rate": 3.484189889518443e-05,
"loss": 0.4212,
"step": 291000
},
{
"epoch": 4.58,
"learning_rate": 3.481561879132547e-05,
"loss": 0.4169,
"step": 291500
},
{
"epoch": 4.59,
"learning_rate": 3.4789391247674214e-05,
"loss": 0.407,
"step": 292000
},
{
"epoch": 4.6,
"learning_rate": 3.476311114381524e-05,
"loss": 0.4112,
"step": 292500
},
{
"epoch": 4.61,
"learning_rate": 3.4736831039956266e-05,
"loss": 0.4177,
"step": 293000
},
{
"epoch": 4.62,
"learning_rate": 3.4710550936097306e-05,
"loss": 0.4272,
"step": 293500
},
{
"epoch": 4.62,
"learning_rate": 3.468427083223833e-05,
"loss": 0.4221,
"step": 294000
},
{
"epoch": 4.63,
"learning_rate": 3.465799072837936e-05,
"loss": 0.4131,
"step": 294500
},
{
"epoch": 4.64,
"learning_rate": 3.463171062452039e-05,
"loss": 0.4238,
"step": 295000
},
{
"epoch": 4.65,
"learning_rate": 3.4605430520661416e-05,
"loss": 0.4213,
"step": 295500
},
{
"epoch": 4.66,
"learning_rate": 3.457915041680245e-05,
"loss": 0.4252,
"step": 296000
},
{
"epoch": 4.66,
"learning_rate": 3.455287031294348e-05,
"loss": 0.4222,
"step": 296500
},
{
"epoch": 4.67,
"learning_rate": 3.452659020908451e-05,
"loss": 0.4206,
"step": 297000
},
{
"epoch": 4.68,
"learning_rate": 3.4500310105225534e-05,
"loss": 0.415,
"step": 297500
},
{
"epoch": 4.69,
"learning_rate": 3.447403000136657e-05,
"loss": 0.4165,
"step": 298000
},
{
"epoch": 4.69,
"learning_rate": 3.44477498975076e-05,
"loss": 0.4266,
"step": 298500
},
{
"epoch": 4.7,
"learning_rate": 3.4421522353856344e-05,
"loss": 0.425,
"step": 299000
},
{
"epoch": 4.71,
"learning_rate": 3.439524224999737e-05,
"loss": 0.42,
"step": 299500
},
{
"epoch": 4.72,
"learning_rate": 3.43689621461384e-05,
"loss": 0.4271,
"step": 300000
},
{
"epoch": 4.73,
"learning_rate": 3.4342734602487155e-05,
"loss": 0.4308,
"step": 300500
},
{
"epoch": 4.73,
"learning_rate": 3.431645449862818e-05,
"loss": 0.4283,
"step": 301000
},
{
"epoch": 4.74,
"learning_rate": 3.429017439476921e-05,
"loss": 0.4296,
"step": 301500
},
{
"epoch": 4.75,
"learning_rate": 3.426389429091024e-05,
"loss": 0.4219,
"step": 302000
},
{
"epoch": 4.76,
"learning_rate": 3.423761418705127e-05,
"loss": 0.4176,
"step": 302500
},
{
"epoch": 4.77,
"learning_rate": 3.42113340831923e-05,
"loss": 0.43,
"step": 303000
},
{
"epoch": 4.77,
"learning_rate": 3.418510653954104e-05,
"loss": 0.4241,
"step": 303500
},
{
"epoch": 4.78,
"learning_rate": 3.4158826435682076e-05,
"loss": 0.4314,
"step": 304000
},
{
"epoch": 4.79,
"learning_rate": 3.413254633182311e-05,
"loss": 0.4217,
"step": 304500
},
{
"epoch": 4.8,
"learning_rate": 3.4106266227964135e-05,
"loss": 0.4378,
"step": 305000
},
{
"epoch": 4.8,
"learning_rate": 3.407998612410516e-05,
"loss": 0.4243,
"step": 305500
},
{
"epoch": 4.81,
"learning_rate": 3.4053706020246193e-05,
"loss": 0.4247,
"step": 306000
},
{
"epoch": 4.82,
"learning_rate": 3.4027425916387226e-05,
"loss": 0.4255,
"step": 306500
},
{
"epoch": 4.83,
"learning_rate": 3.400114581252825e-05,
"loss": 0.4317,
"step": 307000
},
{
"epoch": 4.84,
"learning_rate": 3.3974865708669285e-05,
"loss": 0.4201,
"step": 307500
},
{
"epoch": 4.84,
"learning_rate": 3.394858560481031e-05,
"loss": 0.4273,
"step": 308000
},
{
"epoch": 4.85,
"learning_rate": 3.392230550095134e-05,
"loss": 0.4322,
"step": 308500
},
{
"epoch": 4.86,
"learning_rate": 3.389602539709238e-05,
"loss": 0.4273,
"step": 309000
},
{
"epoch": 4.87,
"learning_rate": 3.386979785344112e-05,
"loss": 0.4291,
"step": 309500
},
{
"epoch": 4.88,
"learning_rate": 3.3843570309789866e-05,
"loss": 0.4234,
"step": 310000
},
{
"epoch": 4.88,
"learning_rate": 3.38172902059309e-05,
"loss": 0.4283,
"step": 310500
},
{
"epoch": 4.89,
"learning_rate": 3.3791010102071925e-05,
"loss": 0.4297,
"step": 311000
},
{
"epoch": 4.9,
"learning_rate": 3.376472999821296e-05,
"loss": 0.4347,
"step": 311500
},
{
"epoch": 4.91,
"learning_rate": 3.37385024545617e-05,
"loss": 0.4325,
"step": 312000
},
{
"epoch": 4.91,
"learning_rate": 3.3712222350702735e-05,
"loss": 0.4215,
"step": 312500
},
{
"epoch": 4.92,
"learning_rate": 3.368594224684376e-05,
"loss": 0.4354,
"step": 313000
},
{
"epoch": 4.93,
"learning_rate": 3.365966214298479e-05,
"loss": 0.4367,
"step": 313500
},
{
"epoch": 4.94,
"learning_rate": 3.363343459933354e-05,
"loss": 0.4402,
"step": 314000
},
{
"epoch": 4.95,
"learning_rate": 3.360715449547457e-05,
"loss": 0.4269,
"step": 314500
},
{
"epoch": 4.95,
"learning_rate": 3.35808743916156e-05,
"loss": 0.433,
"step": 315000
},
{
"epoch": 4.96,
"learning_rate": 3.3554594287756624e-05,
"loss": 0.4356,
"step": 315500
},
{
"epoch": 4.97,
"learning_rate": 3.352831418389766e-05,
"loss": 0.4309,
"step": 316000
},
{
"epoch": 4.98,
"learning_rate": 3.350203408003868e-05,
"loss": 0.4304,
"step": 316500
},
{
"epoch": 4.99,
"learning_rate": 3.3475753976179715e-05,
"loss": 0.433,
"step": 317000
},
{
"epoch": 4.99,
"learning_rate": 3.344947387232075e-05,
"loss": 0.4325,
"step": 317500
},
{
"epoch": 5.0,
"eval_bleu": 56.2388,
"eval_gen_len": 15.235,
"eval_loss": 0.7183424234390259,
"eval_runtime": 9602.8422,
"eval_samples_per_second": 13.443,
"eval_steps_per_second": 1.68,
"step": 317932
},
{
"epoch": 5.0,
"learning_rate": 3.342324632866949e-05,
"loss": 0.418,
"step": 318000
},
{
"epoch": 5.01,
"learning_rate": 3.339696622481052e-05,
"loss": 0.2939,
"step": 318500
},
{
"epoch": 5.02,
"learning_rate": 3.337068612095155e-05,
"loss": 0.2951,
"step": 319000
},
{
"epoch": 5.02,
"learning_rate": 3.3344406017092585e-05,
"loss": 0.2888,
"step": 319500
},
{
"epoch": 5.03,
"learning_rate": 3.331812591323361e-05,
"loss": 0.2897,
"step": 320000
},
{
"epoch": 5.04,
"learning_rate": 3.329184580937464e-05,
"loss": 0.2934,
"step": 320500
},
{
"epoch": 5.05,
"learning_rate": 3.326561826572339e-05,
"loss": 0.2964,
"step": 321000
},
{
"epoch": 5.06,
"learning_rate": 3.323933816186442e-05,
"loss": 0.3002,
"step": 321500
},
{
"epoch": 5.06,
"learning_rate": 3.321305805800545e-05,
"loss": 0.2895,
"step": 322000
},
{
"epoch": 5.07,
"learning_rate": 3.318677795414647e-05,
"loss": 0.3009,
"step": 322500
},
{
"epoch": 5.08,
"learning_rate": 3.3160497850287506e-05,
"loss": 0.2953,
"step": 323000
},
{
"epoch": 5.09,
"learning_rate": 3.313421774642854e-05,
"loss": 0.3046,
"step": 323500
},
{
"epoch": 5.1,
"learning_rate": 3.3108042762985e-05,
"loss": 0.2955,
"step": 324000
},
{
"epoch": 5.1,
"learning_rate": 3.308176265912603e-05,
"loss": 0.3027,
"step": 324500
},
{
"epoch": 5.11,
"learning_rate": 3.305548255526706e-05,
"loss": 0.3,
"step": 325000
},
{
"epoch": 5.12,
"learning_rate": 3.3029202451408094e-05,
"loss": 0.3088,
"step": 325500
},
{
"epoch": 5.13,
"learning_rate": 3.300292234754912e-05,
"loss": 0.2993,
"step": 326000
},
{
"epoch": 5.13,
"learning_rate": 3.2976642243690146e-05,
"loss": 0.3144,
"step": 326500
},
{
"epoch": 5.14,
"learning_rate": 3.29504147000389e-05,
"loss": 0.304,
"step": 327000
},
{
"epoch": 5.15,
"learning_rate": 3.2924134596179923e-05,
"loss": 0.305,
"step": 327500
},
{
"epoch": 5.16,
"learning_rate": 3.2897854492320956e-05,
"loss": 0.2984,
"step": 328000
},
{
"epoch": 5.17,
"learning_rate": 3.287157438846198e-05,
"loss": 0.3084,
"step": 328500
},
{
"epoch": 5.17,
"learning_rate": 3.2845346844810734e-05,
"loss": 0.3135,
"step": 329000
},
{
"epoch": 5.18,
"learning_rate": 3.281906674095176e-05,
"loss": 0.3048,
"step": 329500
},
{
"epoch": 5.19,
"learning_rate": 3.279278663709279e-05,
"loss": 0.3075,
"step": 330000
},
{
"epoch": 5.2,
"learning_rate": 3.276650653323382e-05,
"loss": 0.3074,
"step": 330500
},
{
"epoch": 5.21,
"learning_rate": 3.274022642937485e-05,
"loss": 0.3122,
"step": 331000
},
{
"epoch": 5.21,
"learning_rate": 3.2713946325515884e-05,
"loss": 0.3115,
"step": 331500
},
{
"epoch": 5.22,
"learning_rate": 3.268766622165691e-05,
"loss": 0.3173,
"step": 332000
},
{
"epoch": 5.23,
"learning_rate": 3.2661386117797936e-05,
"loss": 0.3105,
"step": 332500
},
{
"epoch": 5.24,
"learning_rate": 3.263510601393897e-05,
"loss": 0.3154,
"step": 333000
},
{
"epoch": 5.24,
"learning_rate": 3.260887847028772e-05,
"loss": 0.3092,
"step": 333500
},
{
"epoch": 5.25,
"learning_rate": 3.258259836642875e-05,
"loss": 0.3186,
"step": 334000
},
{
"epoch": 5.26,
"learning_rate": 3.255631826256977e-05,
"loss": 0.3146,
"step": 334500
},
{
"epoch": 5.27,
"learning_rate": 3.2530038158710805e-05,
"loss": 0.3163,
"step": 335000
},
{
"epoch": 5.28,
"learning_rate": 3.250375805485184e-05,
"loss": 0.3182,
"step": 335500
},
{
"epoch": 5.28,
"learning_rate": 3.2477477950992864e-05,
"loss": 0.3217,
"step": 336000
},
{
"epoch": 5.29,
"learning_rate": 3.245119784713389e-05,
"loss": 0.3099,
"step": 336500
},
{
"epoch": 5.3,
"learning_rate": 3.242491774327492e-05,
"loss": 0.3191,
"step": 337000
},
{
"epoch": 5.31,
"learning_rate": 3.239863763941595e-05,
"loss": 0.3107,
"step": 337500
},
{
"epoch": 5.32,
"learning_rate": 3.237235753555698e-05,
"loss": 0.3247,
"step": 338000
},
{
"epoch": 5.32,
"learning_rate": 3.2346077431698015e-05,
"loss": 0.3176,
"step": 338500
},
{
"epoch": 5.33,
"learning_rate": 3.231979732783904e-05,
"loss": 0.3249,
"step": 339000
},
{
"epoch": 5.34,
"learning_rate": 3.2293569784187785e-05,
"loss": 0.3186,
"step": 339500
},
{
"epoch": 5.35,
"learning_rate": 3.226728968032882e-05,
"loss": 0.3184,
"step": 340000
},
{
"epoch": 5.35,
"learning_rate": 3.224100957646985e-05,
"loss": 0.32,
"step": 340500
},
{
"epoch": 5.36,
"learning_rate": 3.221472947261088e-05,
"loss": 0.3214,
"step": 341000
},
{
"epoch": 5.37,
"learning_rate": 3.21884493687519e-05,
"loss": 0.3318,
"step": 341500
},
{
"epoch": 5.38,
"learning_rate": 3.2162274385308373e-05,
"loss": 0.3221,
"step": 342000
},
{
"epoch": 5.39,
"learning_rate": 3.21359942814494e-05,
"loss": 0.3179,
"step": 342500
},
{
"epoch": 5.39,
"learning_rate": 3.210971417759043e-05,
"loss": 0.3255,
"step": 343000
},
{
"epoch": 5.4,
"learning_rate": 3.208343407373146e-05,
"loss": 0.3228,
"step": 343500
},
{
"epoch": 5.41,
"learning_rate": 3.205715396987249e-05,
"loss": 0.3269,
"step": 344000
},
{
"epoch": 5.42,
"learning_rate": 3.2030926426221236e-05,
"loss": 0.3273,
"step": 344500
},
{
"epoch": 5.43,
"learning_rate": 3.200469888256998e-05,
"loss": 0.3226,
"step": 345000
},
{
"epoch": 5.43,
"learning_rate": 3.1978418778711013e-05,
"loss": 0.3248,
"step": 345500
},
{
"epoch": 5.44,
"learning_rate": 3.1952138674852046e-05,
"loss": 0.3291,
"step": 346000
},
{
"epoch": 5.45,
"learning_rate": 3.192585857099307e-05,
"loss": 0.3194,
"step": 346500
},
{
"epoch": 5.46,
"learning_rate": 3.1899578467134105e-05,
"loss": 0.3327,
"step": 347000
},
{
"epoch": 5.46,
"learning_rate": 3.187329836327513e-05,
"loss": 0.3264,
"step": 347500
},
{
"epoch": 5.47,
"learning_rate": 3.1847018259416164e-05,
"loss": 0.3339,
"step": 348000
},
{
"epoch": 5.48,
"learning_rate": 3.18207381555572e-05,
"loss": 0.3303,
"step": 348500
},
{
"epoch": 5.49,
"learning_rate": 3.1794563172113653e-05,
"loss": 0.3348,
"step": 349000
},
{
"epoch": 5.5,
"learning_rate": 3.1768283068254686e-05,
"loss": 0.3312,
"step": 349500
},
{
"epoch": 5.5,
"learning_rate": 3.174200296439572e-05,
"loss": 0.3344,
"step": 350000
},
{
"epoch": 5.51,
"learning_rate": 3.1715722860536745e-05,
"loss": 0.3282,
"step": 350500
},
{
"epoch": 5.52,
"learning_rate": 3.168944275667778e-05,
"loss": 0.3316,
"step": 351000
},
{
"epoch": 5.53,
"learning_rate": 3.1663162652818804e-05,
"loss": 0.3269,
"step": 351500
},
{
"epoch": 5.54,
"learning_rate": 3.163688254895984e-05,
"loss": 0.3442,
"step": 352000
},
{
"epoch": 5.54,
"learning_rate": 3.161060244510086e-05,
"loss": 0.3327,
"step": 352500
},
{
"epoch": 5.55,
"learning_rate": 3.1584322341241895e-05,
"loss": 0.3405,
"step": 353000
},
{
"epoch": 5.56,
"learning_rate": 3.155804223738292e-05,
"loss": 0.3369,
"step": 353500
},
{
"epoch": 5.57,
"learning_rate": 3.153181469373167e-05,
"loss": 0.3305,
"step": 354000
},
{
"epoch": 5.58,
"learning_rate": 3.15055345898727e-05,
"loss": 0.3355,
"step": 354500
},
{
"epoch": 5.58,
"learning_rate": 3.147925448601373e-05,
"loss": 0.331,
"step": 355000
},
{
"epoch": 5.59,
"learning_rate": 3.145297438215476e-05,
"loss": 0.3305,
"step": 355500
},
{
"epoch": 5.6,
"learning_rate": 3.142669427829579e-05,
"loss": 0.3289,
"step": 356000
},
{
"epoch": 5.61,
"learning_rate": 3.1400414174436823e-05,
"loss": 0.341,
"step": 356500
},
{
"epoch": 5.61,
"learning_rate": 3.137413407057785e-05,
"loss": 0.3329,
"step": 357000
},
{
"epoch": 5.62,
"learning_rate": 3.1347853966718875e-05,
"loss": 0.3399,
"step": 357500
},
{
"epoch": 5.63,
"learning_rate": 3.132157386285991e-05,
"loss": 0.3305,
"step": 358000
},
{
"epoch": 5.64,
"learning_rate": 3.129529375900094e-05,
"loss": 0.3365,
"step": 358500
},
{
"epoch": 5.65,
"learning_rate": 3.126901365514197e-05,
"loss": 0.3351,
"step": 359000
},
{
"epoch": 5.65,
"learning_rate": 3.1242733551283e-05,
"loss": 0.336,
"step": 359500
},
{
"epoch": 5.66,
"learning_rate": 3.1216506007631745e-05,
"loss": 0.3426,
"step": 360000
},
{
"epoch": 5.67,
"learning_rate": 3.119022590377277e-05,
"loss": 0.3365,
"step": 360500
},
{
"epoch": 5.68,
"learning_rate": 3.1163945799913803e-05,
"loss": 0.3462,
"step": 361000
},
{
"epoch": 5.69,
"learning_rate": 3.113777081647027e-05,
"loss": 0.3325,
"step": 361500
},
{
"epoch": 5.69,
"learning_rate": 3.11114907126113e-05,
"loss": 0.3345,
"step": 362000
},
{
"epoch": 5.7,
"learning_rate": 3.108521060875233e-05,
"loss": 0.3393,
"step": 362500
},
{
"epoch": 5.71,
"learning_rate": 3.105893050489336e-05,
"loss": 0.3354,
"step": 363000
},
{
"epoch": 5.72,
"learning_rate": 3.1032650401034385e-05,
"loss": 0.344,
"step": 363500
},
{
"epoch": 5.72,
"learning_rate": 3.100637029717541e-05,
"loss": 0.337,
"step": 364000
},
{
"epoch": 5.73,
"learning_rate": 3.0980090193316443e-05,
"loss": 0.342,
"step": 364500
},
{
"epoch": 5.74,
"learning_rate": 3.0953810089457476e-05,
"loss": 0.3422,
"step": 365000
},
{
"epoch": 5.75,
"learning_rate": 3.09275299855985e-05,
"loss": 0.3457,
"step": 365500
},
{
"epoch": 5.76,
"learning_rate": 3.0901249881739535e-05,
"loss": 0.345,
"step": 366000
},
{
"epoch": 5.76,
"learning_rate": 3.087502233808828e-05,
"loss": 0.3413,
"step": 366500
},
{
"epoch": 5.77,
"learning_rate": 3.084874223422931e-05,
"loss": 0.3462,
"step": 367000
},
{
"epoch": 5.78,
"learning_rate": 3.082246213037034e-05,
"loss": 0.3434,
"step": 367500
},
{
"epoch": 5.79,
"learning_rate": 3.079618202651137e-05,
"loss": 0.3498,
"step": 368000
},
{
"epoch": 5.8,
"learning_rate": 3.07699019226524e-05,
"loss": 0.3473,
"step": 368500
},
{
"epoch": 5.8,
"learning_rate": 3.074367437900115e-05,
"loss": 0.352,
"step": 369000
},
{
"epoch": 5.81,
"learning_rate": 3.0717394275142175e-05,
"loss": 0.3465,
"step": 369500
},
{
"epoch": 5.82,
"learning_rate": 3.069111417128321e-05,
"loss": 0.3431,
"step": 370000
},
{
"epoch": 5.83,
"learning_rate": 3.0664834067424234e-05,
"loss": 0.3529,
"step": 370500
},
{
"epoch": 5.83,
"learning_rate": 3.0638606523772985e-05,
"loss": 0.3437,
"step": 371000
},
{
"epoch": 5.84,
"learning_rate": 3.061232641991401e-05,
"loss": 0.3498,
"step": 371500
},
{
"epoch": 5.85,
"learning_rate": 3.0586046316055044e-05,
"loss": 0.3525,
"step": 372000
},
{
"epoch": 5.86,
"learning_rate": 3.055976621219607e-05,
"loss": 0.3522,
"step": 372500
},
{
"epoch": 5.87,
"learning_rate": 3.05334861083371e-05,
"loss": 0.3481,
"step": 373000
},
{
"epoch": 5.87,
"learning_rate": 3.0507206004478132e-05,
"loss": 0.3493,
"step": 373500
},
{
"epoch": 5.88,
"learning_rate": 3.0480925900619162e-05,
"loss": 0.3558,
"step": 374000
},
{
"epoch": 5.89,
"learning_rate": 3.045469835696791e-05,
"loss": 0.3498,
"step": 374500
},
{
"epoch": 5.9,
"learning_rate": 3.0428418253108936e-05,
"loss": 0.3445,
"step": 375000
},
{
"epoch": 5.91,
"learning_rate": 3.0402138149249965e-05,
"loss": 0.3541,
"step": 375500
},
{
"epoch": 5.91,
"learning_rate": 3.0375858045390998e-05,
"loss": 0.3462,
"step": 376000
},
{
"epoch": 5.92,
"learning_rate": 3.0349577941532024e-05,
"loss": 0.3411,
"step": 376500
},
{
"epoch": 5.93,
"learning_rate": 3.0323297837673054e-05,
"loss": 0.3471,
"step": 377000
},
{
"epoch": 5.94,
"learning_rate": 3.0297017733814086e-05,
"loss": 0.346,
"step": 377500
},
{
"epoch": 5.94,
"learning_rate": 3.0270737629955116e-05,
"loss": 0.3439,
"step": 378000
},
{
"epoch": 5.95,
"learning_rate": 3.0244562646511583e-05,
"loss": 0.3401,
"step": 378500
},
{
"epoch": 5.96,
"learning_rate": 3.0218335102860328e-05,
"loss": 0.353,
"step": 379000
},
{
"epoch": 5.97,
"learning_rate": 3.0192054999001357e-05,
"loss": 0.358,
"step": 379500
},
{
"epoch": 5.98,
"learning_rate": 3.0165774895142383e-05,
"loss": 0.3475,
"step": 380000
},
{
"epoch": 5.98,
"learning_rate": 3.013949479128342e-05,
"loss": 0.3462,
"step": 380500
},
{
"epoch": 5.99,
"learning_rate": 3.0113214687424445e-05,
"loss": 0.3416,
"step": 381000
},
{
"epoch": 6.0,
"learning_rate": 3.0086934583565475e-05,
"loss": 0.3525,
"step": 381500
},
{
"epoch": 6.0,
"eval_bleu": 55.8867,
"eval_gen_len": 15.2207,
"eval_loss": 0.7727451324462891,
"eval_runtime": 9583.1627,
"eval_samples_per_second": 13.471,
"eval_steps_per_second": 1.684,
"step": 381519
},
{
"epoch": 6.0,
"step": 381519,
"total_flos": 1.3228818910930797e+19,
"train_loss": 0.59605707192661,
"train_runtime": 353473.1733,
"train_samples_per_second": 43.173,
"train_steps_per_second": 2.698
}
],
"logging_steps": 500,
"max_steps": 953790,
"num_input_tokens_seen": 0,
"num_train_epochs": 15,
"save_steps": 500,
"total_flos": 1.3228818910930797e+19,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}