mt5-full-v6.1 / trainer_state.json
phucnn's picture
End of training
2ea33be verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 5000,
"global_step": 107877,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.634779384501298e-08,
"loss": 14.0889,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 9.269558769002596e-08,
"loss": 13.3753,
"step": 200
},
{
"epoch": 0.01,
"learning_rate": 1.3904338153503894e-07,
"loss": 12.3675,
"step": 300
},
{
"epoch": 0.01,
"learning_rate": 1.8539117538005193e-07,
"loss": 11.1815,
"step": 400
},
{
"epoch": 0.01,
"learning_rate": 2.317389692250649e-07,
"loss": 9.3244,
"step": 500
},
{
"epoch": 0.02,
"learning_rate": 2.780867630700779e-07,
"loss": 7.8289,
"step": 600
},
{
"epoch": 0.02,
"learning_rate": 3.2443455691509084e-07,
"loss": 6.5179,
"step": 700
},
{
"epoch": 0.02,
"learning_rate": 3.7078235076010385e-07,
"loss": 5.6367,
"step": 800
},
{
"epoch": 0.03,
"learning_rate": 4.171301446051168e-07,
"loss": 5.0302,
"step": 900
},
{
"epoch": 0.03,
"learning_rate": 4.634779384501298e-07,
"loss": 4.4714,
"step": 1000
},
{
"epoch": 0.03,
"learning_rate": 5.098257322951427e-07,
"loss": 4.0915,
"step": 1100
},
{
"epoch": 0.03,
"learning_rate": 5.561735261401558e-07,
"loss": 3.7279,
"step": 1200
},
{
"epoch": 0.04,
"learning_rate": 6.025213199851688e-07,
"loss": 3.4629,
"step": 1300
},
{
"epoch": 0.04,
"learning_rate": 6.488691138301817e-07,
"loss": 3.2831,
"step": 1400
},
{
"epoch": 0.04,
"learning_rate": 6.952169076751947e-07,
"loss": 3.0667,
"step": 1500
},
{
"epoch": 0.04,
"learning_rate": 7.415647015202077e-07,
"loss": 2.9512,
"step": 1600
},
{
"epoch": 0.05,
"learning_rate": 7.879124953652206e-07,
"loss": 2.7697,
"step": 1700
},
{
"epoch": 0.05,
"learning_rate": 8.342602892102336e-07,
"loss": 2.6436,
"step": 1800
},
{
"epoch": 0.05,
"learning_rate": 8.806080830552465e-07,
"loss": 2.5721,
"step": 1900
},
{
"epoch": 0.06,
"learning_rate": 9.269558769002596e-07,
"loss": 2.547,
"step": 2000
},
{
"epoch": 0.06,
"learning_rate": 9.733036707452726e-07,
"loss": 2.4468,
"step": 2100
},
{
"epoch": 0.06,
"learning_rate": 1.0196514645902855e-06,
"loss": 2.4144,
"step": 2200
},
{
"epoch": 0.06,
"learning_rate": 1.0659992584352986e-06,
"loss": 2.3397,
"step": 2300
},
{
"epoch": 0.07,
"learning_rate": 1.1123470522803115e-06,
"loss": 2.2857,
"step": 2400
},
{
"epoch": 0.07,
"learning_rate": 1.1586948461253246e-06,
"loss": 2.2327,
"step": 2500
},
{
"epoch": 0.07,
"learning_rate": 1.2050426399703375e-06,
"loss": 2.1722,
"step": 2600
},
{
"epoch": 0.08,
"learning_rate": 1.2513904338153504e-06,
"loss": 2.1463,
"step": 2700
},
{
"epoch": 0.08,
"learning_rate": 1.2977382276603633e-06,
"loss": 2.1144,
"step": 2800
},
{
"epoch": 0.08,
"learning_rate": 1.3440860215053765e-06,
"loss": 2.0433,
"step": 2900
},
{
"epoch": 0.08,
"learning_rate": 1.3904338153503894e-06,
"loss": 2.0107,
"step": 3000
},
{
"epoch": 0.09,
"learning_rate": 1.4367816091954023e-06,
"loss": 2.0057,
"step": 3100
},
{
"epoch": 0.09,
"learning_rate": 1.4831294030404154e-06,
"loss": 1.986,
"step": 3200
},
{
"epoch": 0.09,
"learning_rate": 1.5294771968854283e-06,
"loss": 1.9158,
"step": 3300
},
{
"epoch": 0.09,
"learning_rate": 1.5758249907304412e-06,
"loss": 1.9486,
"step": 3400
},
{
"epoch": 0.1,
"learning_rate": 1.6221727845754543e-06,
"loss": 1.85,
"step": 3500
},
{
"epoch": 0.1,
"learning_rate": 1.6685205784204673e-06,
"loss": 1.8479,
"step": 3600
},
{
"epoch": 0.1,
"learning_rate": 1.7148683722654802e-06,
"loss": 1.8052,
"step": 3700
},
{
"epoch": 0.11,
"learning_rate": 1.761216166110493e-06,
"loss": 1.7931,
"step": 3800
},
{
"epoch": 0.11,
"learning_rate": 1.8075639599555062e-06,
"loss": 1.7661,
"step": 3900
},
{
"epoch": 0.11,
"learning_rate": 1.853911753800519e-06,
"loss": 1.7307,
"step": 4000
},
{
"epoch": 0.11,
"learning_rate": 1.900259547645532e-06,
"loss": 1.6742,
"step": 4100
},
{
"epoch": 0.12,
"learning_rate": 1.946607341490545e-06,
"loss": 1.629,
"step": 4200
},
{
"epoch": 0.12,
"learning_rate": 1.9929551353355583e-06,
"loss": 1.6147,
"step": 4300
},
{
"epoch": 0.12,
"learning_rate": 2.039302929180571e-06,
"loss": 1.6556,
"step": 4400
},
{
"epoch": 0.13,
"learning_rate": 2.085650723025584e-06,
"loss": 1.5819,
"step": 4500
},
{
"epoch": 0.13,
"learning_rate": 2.131998516870597e-06,
"loss": 1.6211,
"step": 4600
},
{
"epoch": 0.13,
"learning_rate": 2.17834631071561e-06,
"loss": 1.5386,
"step": 4700
},
{
"epoch": 0.13,
"learning_rate": 2.224694104560623e-06,
"loss": 1.5553,
"step": 4800
},
{
"epoch": 0.14,
"learning_rate": 2.271041898405636e-06,
"loss": 1.5572,
"step": 4900
},
{
"epoch": 0.14,
"learning_rate": 2.3173896922506492e-06,
"loss": 1.5671,
"step": 5000
},
{
"epoch": 0.14,
"eval_bleu": 6.5196,
"eval_gen_len": 18.9699,
"eval_loss": 1.1690529584884644,
"eval_runtime": 968.3873,
"eval_samples_per_second": 2.061,
"eval_steps_per_second": 1.031,
"step": 5000
},
{
"epoch": 0.14,
"learning_rate": 2.3637374860956624e-06,
"loss": 1.4969,
"step": 5100
},
{
"epoch": 0.14,
"learning_rate": 2.410085279940675e-06,
"loss": 1.5547,
"step": 5200
},
{
"epoch": 0.15,
"learning_rate": 2.456433073785688e-06,
"loss": 1.5369,
"step": 5300
},
{
"epoch": 0.15,
"learning_rate": 2.502780867630701e-06,
"loss": 1.4743,
"step": 5400
},
{
"epoch": 0.15,
"learning_rate": 2.549128661475714e-06,
"loss": 1.5065,
"step": 5500
},
{
"epoch": 0.16,
"learning_rate": 2.5954764553207267e-06,
"loss": 1.4637,
"step": 5600
},
{
"epoch": 0.16,
"learning_rate": 2.64182424916574e-06,
"loss": 1.4976,
"step": 5700
},
{
"epoch": 0.16,
"learning_rate": 2.688172043010753e-06,
"loss": 1.482,
"step": 5800
},
{
"epoch": 0.16,
"learning_rate": 2.7345198368557656e-06,
"loss": 1.4703,
"step": 5900
},
{
"epoch": 0.17,
"learning_rate": 2.7808676307007788e-06,
"loss": 1.4279,
"step": 6000
},
{
"epoch": 0.17,
"learning_rate": 2.827215424545792e-06,
"loss": 1.4673,
"step": 6100
},
{
"epoch": 0.17,
"learning_rate": 2.8735632183908046e-06,
"loss": 1.3763,
"step": 6200
},
{
"epoch": 0.18,
"learning_rate": 2.9199110122358177e-06,
"loss": 1.4375,
"step": 6300
},
{
"epoch": 0.18,
"learning_rate": 2.966258806080831e-06,
"loss": 1.3887,
"step": 6400
},
{
"epoch": 0.18,
"learning_rate": 3.0126065999258435e-06,
"loss": 1.3819,
"step": 6500
},
{
"epoch": 0.18,
"learning_rate": 3.0589543937708566e-06,
"loss": 1.4471,
"step": 6600
},
{
"epoch": 0.19,
"learning_rate": 3.1053021876158698e-06,
"loss": 1.3955,
"step": 6700
},
{
"epoch": 0.19,
"learning_rate": 3.1516499814608825e-06,
"loss": 1.4317,
"step": 6800
},
{
"epoch": 0.19,
"learning_rate": 3.1979977753058956e-06,
"loss": 1.3863,
"step": 6900
},
{
"epoch": 0.19,
"learning_rate": 3.2443455691509087e-06,
"loss": 1.3759,
"step": 7000
},
{
"epoch": 0.2,
"learning_rate": 3.2906933629959214e-06,
"loss": 1.3599,
"step": 7100
},
{
"epoch": 0.2,
"learning_rate": 3.3370411568409345e-06,
"loss": 1.4061,
"step": 7200
},
{
"epoch": 0.2,
"learning_rate": 3.383388950685947e-06,
"loss": 1.339,
"step": 7300
},
{
"epoch": 0.21,
"learning_rate": 3.4297367445309603e-06,
"loss": 1.377,
"step": 7400
},
{
"epoch": 0.21,
"learning_rate": 3.4760845383759734e-06,
"loss": 1.3454,
"step": 7500
},
{
"epoch": 0.21,
"learning_rate": 3.522432332220986e-06,
"loss": 1.3565,
"step": 7600
},
{
"epoch": 0.21,
"learning_rate": 3.5687801260659993e-06,
"loss": 1.3381,
"step": 7700
},
{
"epoch": 0.22,
"learning_rate": 3.6151279199110124e-06,
"loss": 1.3326,
"step": 7800
},
{
"epoch": 0.22,
"learning_rate": 3.661475713756025e-06,
"loss": 1.3442,
"step": 7900
},
{
"epoch": 0.22,
"learning_rate": 3.707823507601038e-06,
"loss": 1.3053,
"step": 8000
},
{
"epoch": 0.23,
"learning_rate": 3.7541713014460513e-06,
"loss": 1.3071,
"step": 8100
},
{
"epoch": 0.23,
"learning_rate": 3.800519095291064e-06,
"loss": 1.3401,
"step": 8200
},
{
"epoch": 0.23,
"learning_rate": 3.8468668891360776e-06,
"loss": 1.3061,
"step": 8300
},
{
"epoch": 0.23,
"learning_rate": 3.89321468298109e-06,
"loss": 1.3304,
"step": 8400
},
{
"epoch": 0.24,
"learning_rate": 3.939562476826103e-06,
"loss": 1.3354,
"step": 8500
},
{
"epoch": 0.24,
"learning_rate": 3.9859102706711165e-06,
"loss": 1.3253,
"step": 8600
},
{
"epoch": 0.24,
"learning_rate": 4.032258064516129e-06,
"loss": 1.3006,
"step": 8700
},
{
"epoch": 0.24,
"learning_rate": 4.078605858361142e-06,
"loss": 1.2852,
"step": 8800
},
{
"epoch": 0.25,
"learning_rate": 4.1249536522061554e-06,
"loss": 1.2775,
"step": 8900
},
{
"epoch": 0.25,
"learning_rate": 4.171301446051168e-06,
"loss": 1.2842,
"step": 9000
},
{
"epoch": 0.25,
"learning_rate": 4.217649239896181e-06,
"loss": 1.2935,
"step": 9100
},
{
"epoch": 0.26,
"learning_rate": 4.263997033741194e-06,
"loss": 1.2656,
"step": 9200
},
{
"epoch": 0.26,
"learning_rate": 4.310344827586207e-06,
"loss": 1.27,
"step": 9300
},
{
"epoch": 0.26,
"learning_rate": 4.35669262143122e-06,
"loss": 1.2983,
"step": 9400
},
{
"epoch": 0.26,
"learning_rate": 4.403040415276233e-06,
"loss": 1.2317,
"step": 9500
},
{
"epoch": 0.27,
"learning_rate": 4.449388209121246e-06,
"loss": 1.2491,
"step": 9600
},
{
"epoch": 0.27,
"learning_rate": 4.4957360029662596e-06,
"loss": 1.2438,
"step": 9700
},
{
"epoch": 0.27,
"learning_rate": 4.542083796811272e-06,
"loss": 1.2608,
"step": 9800
},
{
"epoch": 0.28,
"learning_rate": 4.588431590656286e-06,
"loss": 1.2656,
"step": 9900
},
{
"epoch": 0.28,
"learning_rate": 4.6347793845012985e-06,
"loss": 1.2277,
"step": 10000
},
{
"epoch": 0.28,
"eval_bleu": 7.082,
"eval_gen_len": 18.9724,
"eval_loss": 1.0592412948608398,
"eval_runtime": 961.2611,
"eval_samples_per_second": 2.076,
"eval_steps_per_second": 1.038,
"step": 10000
},
{
"epoch": 0.28,
"learning_rate": 4.681127178346311e-06,
"loss": 1.3006,
"step": 10100
},
{
"epoch": 0.28,
"learning_rate": 4.727474972191325e-06,
"loss": 1.2365,
"step": 10200
},
{
"epoch": 0.29,
"learning_rate": 4.7738227660363374e-06,
"loss": 1.252,
"step": 10300
},
{
"epoch": 0.29,
"learning_rate": 4.82017055988135e-06,
"loss": 1.2351,
"step": 10400
},
{
"epoch": 0.29,
"learning_rate": 4.866518353726364e-06,
"loss": 1.2442,
"step": 10500
},
{
"epoch": 0.29,
"learning_rate": 4.912866147571376e-06,
"loss": 1.2158,
"step": 10600
},
{
"epoch": 0.3,
"learning_rate": 4.959213941416389e-06,
"loss": 1.2123,
"step": 10700
},
{
"epoch": 0.3,
"learning_rate": 4.999999811534358e-06,
"loss": 1.1988,
"step": 10800
},
{
"epoch": 0.3,
"learning_rate": 4.999983582566248e-06,
"loss": 1.2438,
"step": 10900
},
{
"epoch": 0.31,
"learning_rate": 4.999941178008878e-06,
"loss": 1.2145,
"step": 11000
},
{
"epoch": 0.31,
"learning_rate": 4.999872598306237e-06,
"loss": 1.1855,
"step": 11100
},
{
"epoch": 0.31,
"learning_rate": 4.999777844176376e-06,
"loss": 1.2148,
"step": 11200
},
{
"epoch": 0.31,
"learning_rate": 4.999656916611398e-06,
"loss": 1.2128,
"step": 11300
},
{
"epoch": 0.32,
"learning_rate": 4.999509816877453e-06,
"loss": 1.2222,
"step": 11400
},
{
"epoch": 0.32,
"learning_rate": 4.999336546514719e-06,
"loss": 1.1951,
"step": 11500
},
{
"epoch": 0.32,
"learning_rate": 4.9991371073373895e-06,
"loss": 1.1656,
"step": 11600
},
{
"epoch": 0.33,
"learning_rate": 4.998911501433653e-06,
"loss": 1.1783,
"step": 11700
},
{
"epoch": 0.33,
"learning_rate": 4.9986597311656735e-06,
"loss": 1.182,
"step": 11800
},
{
"epoch": 0.33,
"learning_rate": 4.998381799169562e-06,
"loss": 1.1869,
"step": 11900
},
{
"epoch": 0.33,
"learning_rate": 4.99807770835535e-06,
"loss": 1.2242,
"step": 12000
},
{
"epoch": 0.34,
"learning_rate": 4.997747461906961e-06,
"loss": 1.1884,
"step": 12100
},
{
"epoch": 0.34,
"learning_rate": 4.997391063282177e-06,
"loss": 1.2025,
"step": 12200
},
{
"epoch": 0.34,
"learning_rate": 4.997008516212599e-06,
"loss": 1.207,
"step": 12300
},
{
"epoch": 0.34,
"learning_rate": 4.996599824703613e-06,
"loss": 1.1898,
"step": 12400
},
{
"epoch": 0.35,
"learning_rate": 4.996164993034341e-06,
"loss": 1.1942,
"step": 12500
},
{
"epoch": 0.35,
"learning_rate": 4.995704025757605e-06,
"loss": 1.1647,
"step": 12600
},
{
"epoch": 0.35,
"learning_rate": 4.995216927699872e-06,
"loss": 1.1961,
"step": 12700
},
{
"epoch": 0.36,
"learning_rate": 4.994703703961206e-06,
"loss": 1.1728,
"step": 12800
},
{
"epoch": 0.36,
"learning_rate": 4.994164359915219e-06,
"loss": 1.1801,
"step": 12900
},
{
"epoch": 0.36,
"learning_rate": 4.993598901209003e-06,
"loss": 1.1607,
"step": 13000
},
{
"epoch": 0.36,
"learning_rate": 4.993007333763086e-06,
"loss": 1.1674,
"step": 13100
},
{
"epoch": 0.37,
"learning_rate": 4.9923896637713575e-06,
"loss": 1.1837,
"step": 13200
},
{
"epoch": 0.37,
"learning_rate": 4.991745897701012e-06,
"loss": 1.1709,
"step": 13300
},
{
"epoch": 0.37,
"learning_rate": 4.991076042292475e-06,
"loss": 1.1672,
"step": 13400
},
{
"epoch": 0.38,
"learning_rate": 4.990380104559337e-06,
"loss": 1.1649,
"step": 13500
},
{
"epoch": 0.38,
"learning_rate": 4.989658091788277e-06,
"loss": 1.1582,
"step": 13600
},
{
"epoch": 0.38,
"learning_rate": 4.988910011538991e-06,
"loss": 1.1485,
"step": 13700
},
{
"epoch": 0.38,
"learning_rate": 4.988135871644105e-06,
"loss": 1.1887,
"step": 13800
},
{
"epoch": 0.39,
"learning_rate": 4.987335680209099e-06,
"loss": 1.1656,
"step": 13900
},
{
"epoch": 0.39,
"learning_rate": 4.986509445612223e-06,
"loss": 1.1182,
"step": 14000
},
{
"epoch": 0.39,
"learning_rate": 4.985657176504402e-06,
"loss": 1.0969,
"step": 14100
},
{
"epoch": 0.39,
"learning_rate": 4.984778881809156e-06,
"loss": 1.1648,
"step": 14200
},
{
"epoch": 0.4,
"learning_rate": 4.983874570722496e-06,
"loss": 1.1689,
"step": 14300
},
{
"epoch": 0.4,
"learning_rate": 4.982944252712834e-06,
"loss": 1.1127,
"step": 14400
},
{
"epoch": 0.4,
"learning_rate": 4.981987937520884e-06,
"loss": 1.1529,
"step": 14500
},
{
"epoch": 0.41,
"learning_rate": 4.981005635159558e-06,
"loss": 1.1279,
"step": 14600
},
{
"epoch": 0.41,
"learning_rate": 4.979997355913859e-06,
"loss": 1.1099,
"step": 14700
},
{
"epoch": 0.41,
"learning_rate": 4.978963110340778e-06,
"loss": 1.1671,
"step": 14800
},
{
"epoch": 0.41,
"learning_rate": 4.9779029092691825e-06,
"loss": 1.167,
"step": 14900
},
{
"epoch": 0.42,
"learning_rate": 4.976816763799698e-06,
"loss": 1.1316,
"step": 15000
},
{
"epoch": 0.42,
"eval_bleu": 7.3283,
"eval_gen_len": 18.9825,
"eval_loss": 1.0111815929412842,
"eval_runtime": 963.287,
"eval_samples_per_second": 2.072,
"eval_steps_per_second": 1.036,
"step": 15000
},
{
"epoch": 0.42,
"learning_rate": 4.975704685304601e-06,
"loss": 1.1448,
"step": 15100
},
{
"epoch": 0.42,
"learning_rate": 4.97456668542769e-06,
"loss": 1.1221,
"step": 15200
},
{
"epoch": 0.43,
"learning_rate": 4.97340277608417e-06,
"loss": 1.1549,
"step": 15300
},
{
"epoch": 0.43,
"learning_rate": 4.972212969460528e-06,
"loss": 1.1493,
"step": 15400
},
{
"epoch": 0.43,
"learning_rate": 4.9709972780144e-06,
"loss": 1.1595,
"step": 15500
},
{
"epoch": 0.43,
"learning_rate": 4.969755714474447e-06,
"loss": 1.1123,
"step": 15600
},
{
"epoch": 0.44,
"learning_rate": 4.968488291840215e-06,
"loss": 1.123,
"step": 15700
},
{
"epoch": 0.44,
"learning_rate": 4.9671950233820075e-06,
"loss": 1.1096,
"step": 15800
},
{
"epoch": 0.44,
"learning_rate": 4.965875922640738e-06,
"loss": 1.1296,
"step": 15900
},
{
"epoch": 0.44,
"learning_rate": 4.964531003427792e-06,
"loss": 1.1528,
"step": 16000
},
{
"epoch": 0.45,
"learning_rate": 4.9631602798248845e-06,
"loss": 1.1315,
"step": 16100
},
{
"epoch": 0.45,
"learning_rate": 4.961763766183908e-06,
"loss": 1.0704,
"step": 16200
},
{
"epoch": 0.45,
"learning_rate": 4.960341477126786e-06,
"loss": 1.1107,
"step": 16300
},
{
"epoch": 0.46,
"learning_rate": 4.9588934275453165e-06,
"loss": 1.1351,
"step": 16400
},
{
"epoch": 0.46,
"learning_rate": 4.957419632601022e-06,
"loss": 1.1132,
"step": 16500
},
{
"epoch": 0.46,
"learning_rate": 4.955920107724982e-06,
"loss": 1.0942,
"step": 16600
},
{
"epoch": 0.46,
"learning_rate": 4.95439486861768e-06,
"loss": 1.1466,
"step": 16700
},
{
"epoch": 0.47,
"learning_rate": 4.952843931248834e-06,
"loss": 1.1164,
"step": 16800
},
{
"epoch": 0.47,
"learning_rate": 4.951267311857229e-06,
"loss": 1.1336,
"step": 16900
},
{
"epoch": 0.47,
"learning_rate": 4.949665026950551e-06,
"loss": 1.0946,
"step": 17000
},
{
"epoch": 0.48,
"learning_rate": 4.948037093305211e-06,
"loss": 1.0768,
"step": 17100
},
{
"epoch": 0.48,
"learning_rate": 4.94638352796617e-06,
"loss": 1.1048,
"step": 17200
},
{
"epoch": 0.48,
"learning_rate": 4.944704348246759e-06,
"loss": 1.0964,
"step": 17300
},
{
"epoch": 0.48,
"learning_rate": 4.942999571728503e-06,
"loss": 1.115,
"step": 17400
},
{
"epoch": 0.49,
"learning_rate": 4.941269216260929e-06,
"loss": 1.0912,
"step": 17500
},
{
"epoch": 0.49,
"learning_rate": 4.9395132999613874e-06,
"loss": 1.1039,
"step": 17600
},
{
"epoch": 0.49,
"learning_rate": 4.937731841214856e-06,
"loss": 1.1001,
"step": 17700
},
{
"epoch": 0.5,
"learning_rate": 4.935924858673751e-06,
"loss": 1.1137,
"step": 17800
},
{
"epoch": 0.5,
"learning_rate": 4.934092371257727e-06,
"loss": 1.1154,
"step": 17900
},
{
"epoch": 0.5,
"learning_rate": 4.932234398153488e-06,
"loss": 1.1103,
"step": 18000
},
{
"epoch": 0.5,
"learning_rate": 4.930350958814578e-06,
"loss": 1.0984,
"step": 18100
},
{
"epoch": 0.51,
"learning_rate": 4.9284420729611785e-06,
"loss": 1.1006,
"step": 18200
},
{
"epoch": 0.51,
"learning_rate": 4.926507760579906e-06,
"loss": 1.1028,
"step": 18300
},
{
"epoch": 0.51,
"learning_rate": 4.9245480419236015e-06,
"loss": 1.0915,
"step": 18400
},
{
"epoch": 0.51,
"learning_rate": 4.922562937511115e-06,
"loss": 1.0716,
"step": 18500
},
{
"epoch": 0.52,
"learning_rate": 4.920552468127093e-06,
"loss": 1.0817,
"step": 18600
},
{
"epoch": 0.52,
"learning_rate": 4.918516654821765e-06,
"loss": 1.113,
"step": 18700
},
{
"epoch": 0.52,
"learning_rate": 4.916455518910713e-06,
"loss": 1.1086,
"step": 18800
},
{
"epoch": 0.53,
"learning_rate": 4.9143690819746595e-06,
"loss": 1.0853,
"step": 18900
},
{
"epoch": 0.53,
"learning_rate": 4.912257365859234e-06,
"loss": 1.1003,
"step": 19000
},
{
"epoch": 0.53,
"learning_rate": 4.9101203926747465e-06,
"loss": 1.074,
"step": 19100
},
{
"epoch": 0.53,
"learning_rate": 4.907958184795958e-06,
"loss": 1.069,
"step": 19200
},
{
"epoch": 0.54,
"learning_rate": 4.905770764861842e-06,
"loss": 1.0779,
"step": 19300
},
{
"epoch": 0.54,
"learning_rate": 4.903558155775352e-06,
"loss": 1.0746,
"step": 19400
},
{
"epoch": 0.54,
"learning_rate": 4.901320380703179e-06,
"loss": 1.0855,
"step": 19500
},
{
"epoch": 0.55,
"learning_rate": 4.8990574630755085e-06,
"loss": 1.0499,
"step": 19600
},
{
"epoch": 0.55,
"learning_rate": 4.896769426585778e-06,
"loss": 1.0826,
"step": 19700
},
{
"epoch": 0.55,
"learning_rate": 4.8944562951904256e-06,
"loss": 1.0936,
"step": 19800
},
{
"epoch": 0.55,
"learning_rate": 4.892118093108641e-06,
"loss": 1.0893,
"step": 19900
},
{
"epoch": 0.56,
"learning_rate": 4.889754844822113e-06,
"loss": 1.0833,
"step": 20000
},
{
"epoch": 0.56,
"eval_bleu": 7.4462,
"eval_gen_len": 18.977,
"eval_loss": 0.972794771194458,
"eval_runtime": 963.5918,
"eval_samples_per_second": 2.071,
"eval_steps_per_second": 1.036,
"step": 20000
},
{
"epoch": 0.56,
"learning_rate": 4.887366575074769e-06,
"loss": 1.1106,
"step": 20100
},
{
"epoch": 0.56,
"learning_rate": 4.884953308872522e-06,
"loss": 1.0961,
"step": 20200
},
{
"epoch": 0.56,
"learning_rate": 4.882515071483003e-06,
"loss": 1.0668,
"step": 20300
},
{
"epoch": 0.57,
"learning_rate": 4.8800518884353e-06,
"loss": 1.0548,
"step": 20400
},
{
"epoch": 0.57,
"learning_rate": 4.8775637855196885e-06,
"loss": 1.1031,
"step": 20500
},
{
"epoch": 0.57,
"learning_rate": 4.875050788787367e-06,
"loss": 1.0564,
"step": 20600
},
{
"epoch": 0.58,
"learning_rate": 4.872512924550172e-06,
"loss": 1.0837,
"step": 20700
},
{
"epoch": 0.58,
"learning_rate": 4.869950219380317e-06,
"loss": 1.0816,
"step": 20800
},
{
"epoch": 0.58,
"learning_rate": 4.867362700110105e-06,
"loss": 1.0352,
"step": 20900
},
{
"epoch": 0.58,
"learning_rate": 4.864750393831651e-06,
"loss": 1.068,
"step": 21000
},
{
"epoch": 0.59,
"learning_rate": 4.8621133278965956e-06,
"loss": 1.0525,
"step": 21100
},
{
"epoch": 0.59,
"learning_rate": 4.859451529915825e-06,
"loss": 1.0867,
"step": 21200
},
{
"epoch": 0.59,
"learning_rate": 4.856765027759171e-06,
"loss": 1.0544,
"step": 21300
},
{
"epoch": 0.6,
"learning_rate": 4.8540538495551314e-06,
"loss": 1.0805,
"step": 21400
},
{
"epoch": 0.6,
"learning_rate": 4.851318023690567e-06,
"loss": 1.0598,
"step": 21500
},
{
"epoch": 0.6,
"learning_rate": 4.848557578810407e-06,
"loss": 1.0418,
"step": 21600
},
{
"epoch": 0.6,
"learning_rate": 4.845772543817351e-06,
"loss": 1.0486,
"step": 21700
},
{
"epoch": 0.61,
"learning_rate": 4.842962947871561e-06,
"loss": 1.0618,
"step": 21800
},
{
"epoch": 0.61,
"learning_rate": 4.840128820390364e-06,
"loss": 1.0583,
"step": 21900
},
{
"epoch": 0.61,
"learning_rate": 4.837270191047937e-06,
"loss": 1.0703,
"step": 22000
},
{
"epoch": 0.61,
"learning_rate": 4.834387089774999e-06,
"loss": 1.039,
"step": 22100
},
{
"epoch": 0.62,
"learning_rate": 4.8314795467585e-06,
"loss": 1.0505,
"step": 22200
},
{
"epoch": 0.62,
"learning_rate": 4.828547592441298e-06,
"loss": 1.0539,
"step": 22300
},
{
"epoch": 0.62,
"learning_rate": 4.8255912575218485e-06,
"loss": 1.0375,
"step": 22400
},
{
"epoch": 0.63,
"learning_rate": 4.8226105729538786e-06,
"loss": 1.0378,
"step": 22500
},
{
"epoch": 0.63,
"learning_rate": 4.8196055699460636e-06,
"loss": 1.0304,
"step": 22600
},
{
"epoch": 0.63,
"learning_rate": 4.816576279961699e-06,
"loss": 1.065,
"step": 22700
},
{
"epoch": 0.63,
"learning_rate": 4.813522734718373e-06,
"loss": 1.0445,
"step": 22800
},
{
"epoch": 0.64,
"learning_rate": 4.810444966187635e-06,
"loss": 1.0327,
"step": 22900
},
{
"epoch": 0.64,
"learning_rate": 4.807343006594658e-06,
"loss": 1.0751,
"step": 23000
},
{
"epoch": 0.64,
"learning_rate": 4.804216888417904e-06,
"loss": 1.0444,
"step": 23100
},
{
"epoch": 0.65,
"learning_rate": 4.801066644388781e-06,
"loss": 1.005,
"step": 23200
},
{
"epoch": 0.65,
"learning_rate": 4.797892307491303e-06,
"loss": 1.0401,
"step": 23300
},
{
"epoch": 0.65,
"learning_rate": 4.794693910961745e-06,
"loss": 1.0566,
"step": 23400
},
{
"epoch": 0.65,
"learning_rate": 4.7914714882882924e-06,
"loss": 1.0478,
"step": 23500
},
{
"epoch": 0.66,
"learning_rate": 4.788225073210691e-06,
"loss": 1.0656,
"step": 23600
},
{
"epoch": 0.66,
"learning_rate": 4.784954699719895e-06,
"loss": 1.0244,
"step": 23700
},
{
"epoch": 0.66,
"learning_rate": 4.7816604020577105e-06,
"loss": 1.0288,
"step": 23800
},
{
"epoch": 0.66,
"learning_rate": 4.778342214716438e-06,
"loss": 1.0515,
"step": 23900
},
{
"epoch": 0.67,
"learning_rate": 4.775000172438508e-06,
"loss": 1.0391,
"step": 24000
},
{
"epoch": 0.67,
"learning_rate": 4.771634310216122e-06,
"loss": 1.0481,
"step": 24100
},
{
"epoch": 0.67,
"learning_rate": 4.768244663290881e-06,
"loss": 1.0502,
"step": 24200
},
{
"epoch": 0.68,
"learning_rate": 4.76483126715342e-06,
"loss": 1.0855,
"step": 24300
},
{
"epoch": 0.68,
"learning_rate": 4.761394157543038e-06,
"loss": 1.0305,
"step": 24400
},
{
"epoch": 0.68,
"learning_rate": 4.757933370447317e-06,
"loss": 1.0609,
"step": 24500
},
{
"epoch": 0.68,
"learning_rate": 4.754448942101753e-06,
"loss": 1.0152,
"step": 24600
},
{
"epoch": 0.69,
"learning_rate": 4.7509409089893695e-06,
"loss": 1.0449,
"step": 24700
},
{
"epoch": 0.69,
"learning_rate": 4.7474093078403436e-06,
"loss": 1.0152,
"step": 24800
},
{
"epoch": 0.69,
"learning_rate": 4.743854175631614e-06,
"loss": 1.0,
"step": 24900
},
{
"epoch": 0.7,
"learning_rate": 4.740275549586496e-06,
"loss": 1.0339,
"step": 25000
},
{
"epoch": 0.7,
"eval_bleu": 8.0126,
"eval_gen_len": 18.982,
"eval_loss": 0.9545806646347046,
"eval_runtime": 963.2133,
"eval_samples_per_second": 2.072,
"eval_steps_per_second": 1.036,
"step": 25000
},
{
"epoch": 0.7,
"learning_rate": 4.736673467174295e-06,
"loss": 1.0262,
"step": 25100
},
{
"epoch": 0.7,
"learning_rate": 4.733047966109911e-06,
"loss": 1.054,
"step": 25200
},
{
"epoch": 0.7,
"learning_rate": 4.729399084353444e-06,
"loss": 0.9769,
"step": 25300
},
{
"epoch": 0.71,
"learning_rate": 4.725726860109794e-06,
"loss": 1.0007,
"step": 25400
},
{
"epoch": 0.71,
"learning_rate": 4.7220313318282704e-06,
"loss": 1.0438,
"step": 25500
},
{
"epoch": 0.71,
"learning_rate": 4.718312538202179e-06,
"loss": 1.0347,
"step": 25600
},
{
"epoch": 0.71,
"learning_rate": 4.7145705181684195e-06,
"loss": 1.04,
"step": 25700
},
{
"epoch": 0.72,
"learning_rate": 4.710805310907083e-06,
"loss": 1.0099,
"step": 25800
},
{
"epoch": 0.72,
"learning_rate": 4.707016955841034e-06,
"loss": 1.0447,
"step": 25900
},
{
"epoch": 0.72,
"learning_rate": 4.703205492635504e-06,
"loss": 0.9819,
"step": 26000
},
{
"epoch": 0.73,
"learning_rate": 4.699370961197675e-06,
"loss": 1.0066,
"step": 26100
},
{
"epoch": 0.73,
"learning_rate": 4.695513401676256e-06,
"loss": 1.007,
"step": 26200
},
{
"epoch": 0.73,
"learning_rate": 4.691632854461071e-06,
"loss": 1.037,
"step": 26300
},
{
"epoch": 0.73,
"learning_rate": 4.68772936018263e-06,
"loss": 1.0054,
"step": 26400
},
{
"epoch": 0.74,
"learning_rate": 4.683802959711709e-06,
"loss": 1.0078,
"step": 26500
},
{
"epoch": 0.74,
"learning_rate": 4.6798536941589125e-06,
"loss": 1.0187,
"step": 26600
},
{
"epoch": 0.74,
"learning_rate": 4.675881604874257e-06,
"loss": 1.0209,
"step": 26700
},
{
"epoch": 0.75,
"learning_rate": 4.6718867334467245e-06,
"loss": 0.9908,
"step": 26800
},
{
"epoch": 0.75,
"learning_rate": 4.667869121703835e-06,
"loss": 0.9868,
"step": 26900
},
{
"epoch": 0.75,
"learning_rate": 4.6638288117112075e-06,
"loss": 0.985,
"step": 27000
},
{
"epoch": 0.75,
"learning_rate": 4.6597658457721175e-06,
"loss": 1.0143,
"step": 27100
},
{
"epoch": 0.76,
"learning_rate": 4.655680266427057e-06,
"loss": 0.9698,
"step": 27200
},
{
"epoch": 0.76,
"learning_rate": 4.651572116453281e-06,
"loss": 1.0149,
"step": 27300
},
{
"epoch": 0.76,
"learning_rate": 4.6474414388643755e-06,
"loss": 1.0212,
"step": 27400
},
{
"epoch": 0.76,
"learning_rate": 4.643288276909791e-06,
"loss": 1.0126,
"step": 27500
},
{
"epoch": 0.77,
"learning_rate": 4.639112674074396e-06,
"loss": 1.0163,
"step": 27600
},
{
"epoch": 0.77,
"learning_rate": 4.634914674078025e-06,
"loss": 0.9814,
"step": 27700
},
{
"epoch": 0.77,
"learning_rate": 4.630694320875016e-06,
"loss": 1.039,
"step": 27800
},
{
"epoch": 0.78,
"learning_rate": 4.626451658653752e-06,
"loss": 1.0087,
"step": 27900
},
{
"epoch": 0.78,
"learning_rate": 4.6221867318361975e-06,
"loss": 0.9795,
"step": 28000
},
{
"epoch": 0.78,
"learning_rate": 4.617899585077436e-06,
"loss": 1.0499,
"step": 28100
},
{
"epoch": 0.78,
"learning_rate": 4.613590263265198e-06,
"loss": 1.0109,
"step": 28200
},
{
"epoch": 0.79,
"learning_rate": 4.6092588115193945e-06,
"loss": 1.0347,
"step": 28300
},
{
"epoch": 0.79,
"learning_rate": 4.604905275191647e-06,
"loss": 1.0123,
"step": 28400
},
{
"epoch": 0.79,
"learning_rate": 4.600529699864803e-06,
"loss": 1.0216,
"step": 28500
},
{
"epoch": 0.8,
"learning_rate": 4.59613213135247e-06,
"loss": 0.9955,
"step": 28600
},
{
"epoch": 0.8,
"learning_rate": 4.59171261569853e-06,
"loss": 0.9794,
"step": 28700
},
{
"epoch": 0.8,
"learning_rate": 4.587271199176654e-06,
"loss": 1.0098,
"step": 28800
},
{
"epoch": 0.8,
"learning_rate": 4.58280792828983e-06,
"loss": 1.0176,
"step": 28900
},
{
"epoch": 0.81,
"learning_rate": 4.57832284976986e-06,
"loss": 1.0104,
"step": 29000
},
{
"epoch": 0.81,
"learning_rate": 4.5738160105768815e-06,
"loss": 1.0023,
"step": 29100
},
{
"epoch": 0.81,
"learning_rate": 4.569287457898874e-06,
"loss": 1.0333,
"step": 29200
},
{
"epoch": 0.81,
"learning_rate": 4.564737239151164e-06,
"loss": 0.9876,
"step": 29300
},
{
"epoch": 0.82,
"learning_rate": 4.560165401975925e-06,
"loss": 0.9966,
"step": 29400
},
{
"epoch": 0.82,
"learning_rate": 4.555571994241685e-06,
"loss": 1.0267,
"step": 29500
},
{
"epoch": 0.82,
"learning_rate": 4.550957064042821e-06,
"loss": 1.0371,
"step": 29600
},
{
"epoch": 0.83,
"learning_rate": 4.546320659699059e-06,
"loss": 0.9998,
"step": 29700
},
{
"epoch": 0.83,
"learning_rate": 4.541662829754963e-06,
"loss": 1.0174,
"step": 29800
},
{
"epoch": 0.83,
"learning_rate": 4.536983622979429e-06,
"loss": 0.99,
"step": 29900
},
{
"epoch": 0.83,
"learning_rate": 4.532283088365179e-06,
"loss": 1.025,
"step": 30000
},
{
"epoch": 0.83,
"eval_bleu": 7.7648,
"eval_gen_len": 18.9805,
"eval_loss": 0.9337242245674133,
"eval_runtime": 963.8615,
"eval_samples_per_second": 2.071,
"eval_steps_per_second": 1.035,
"step": 30000
},
{
"epoch": 0.84,
"learning_rate": 4.527561275128241e-06,
"loss": 1.03,
"step": 30100
},
{
"epoch": 0.84,
"learning_rate": 4.522818232707435e-06,
"loss": 1.0015,
"step": 30200
},
{
"epoch": 0.84,
"learning_rate": 4.5180540107638634e-06,
"loss": 0.9622,
"step": 30300
},
{
"epoch": 0.85,
"learning_rate": 4.513268659180377e-06,
"loss": 1.0272,
"step": 30400
},
{
"epoch": 0.85,
"learning_rate": 4.508462228061065e-06,
"loss": 0.994,
"step": 30500
},
{
"epoch": 0.85,
"learning_rate": 4.503634767730724e-06,
"loss": 0.9826,
"step": 30600
},
{
"epoch": 0.85,
"learning_rate": 4.498786328734336e-06,
"loss": 0.9823,
"step": 30700
},
{
"epoch": 0.86,
"learning_rate": 4.49391696183653e-06,
"loss": 0.9975,
"step": 30800
},
{
"epoch": 0.86,
"learning_rate": 4.489026718021061e-06,
"loss": 0.9776,
"step": 30900
},
{
"epoch": 0.86,
"learning_rate": 4.484115648490271e-06,
"loss": 0.9798,
"step": 31000
},
{
"epoch": 0.86,
"learning_rate": 4.4791838046645545e-06,
"loss": 0.9634,
"step": 31100
},
{
"epoch": 0.87,
"learning_rate": 4.474231238181817e-06,
"loss": 1.0174,
"step": 31200
},
{
"epoch": 0.87,
"learning_rate": 4.469258000896936e-06,
"loss": 0.9762,
"step": 31300
},
{
"epoch": 0.87,
"learning_rate": 4.464264144881221e-06,
"loss": 1.0287,
"step": 31400
},
{
"epoch": 0.88,
"learning_rate": 4.459249722421866e-06,
"loss": 0.977,
"step": 31500
},
{
"epoch": 0.88,
"learning_rate": 4.454214786021399e-06,
"loss": 0.9885,
"step": 31600
},
{
"epoch": 0.88,
"learning_rate": 4.449159388397138e-06,
"loss": 1.0286,
"step": 31700
},
{
"epoch": 0.88,
"learning_rate": 4.4440835824806364e-06,
"loss": 0.9907,
"step": 31800
},
{
"epoch": 0.89,
"learning_rate": 4.438987421417126e-06,
"loss": 0.9827,
"step": 31900
},
{
"epoch": 0.89,
"learning_rate": 4.433870958564965e-06,
"loss": 0.9911,
"step": 32000
},
{
"epoch": 0.89,
"learning_rate": 4.428734247495077e-06,
"loss": 1.0034,
"step": 32100
},
{
"epoch": 0.9,
"learning_rate": 4.423577341990392e-06,
"loss": 0.9539,
"step": 32200
},
{
"epoch": 0.9,
"learning_rate": 4.41840029604528e-06,
"loss": 1.0161,
"step": 32300
},
{
"epoch": 0.9,
"learning_rate": 4.413203163864988e-06,
"loss": 1.0044,
"step": 32400
},
{
"epoch": 0.9,
"learning_rate": 4.407985999865072e-06,
"loss": 0.9767,
"step": 32500
},
{
"epoch": 0.91,
"learning_rate": 4.4027488586708274e-06,
"loss": 0.9824,
"step": 32600
},
{
"epoch": 0.91,
"learning_rate": 4.397491795116719e-06,
"loss": 0.9905,
"step": 32700
},
{
"epoch": 0.91,
"learning_rate": 4.392214864245801e-06,
"loss": 0.9632,
"step": 32800
},
{
"epoch": 0.91,
"learning_rate": 4.386918121309147e-06,
"loss": 0.9724,
"step": 32900
},
{
"epoch": 0.92,
"learning_rate": 4.381601621765267e-06,
"loss": 1.0322,
"step": 33000
},
{
"epoch": 0.92,
"learning_rate": 4.376265421279532e-06,
"loss": 1.0209,
"step": 33100
},
{
"epoch": 0.92,
"learning_rate": 4.3709095757235835e-06,
"loss": 0.9452,
"step": 33200
},
{
"epoch": 0.93,
"learning_rate": 4.365534141174756e-06,
"loss": 0.9999,
"step": 33300
},
{
"epoch": 0.93,
"learning_rate": 4.360139173915486e-06,
"loss": 0.9786,
"step": 33400
},
{
"epoch": 0.93,
"learning_rate": 4.3547247304327234e-06,
"loss": 1.0074,
"step": 33500
},
{
"epoch": 0.93,
"learning_rate": 4.34929086741734e-06,
"loss": 0.9756,
"step": 33600
},
{
"epoch": 0.94,
"learning_rate": 4.343837641763535e-06,
"loss": 0.9844,
"step": 33700
},
{
"epoch": 0.94,
"learning_rate": 4.338365110568242e-06,
"loss": 1.0181,
"step": 33800
},
{
"epoch": 0.94,
"learning_rate": 4.332873331130531e-06,
"loss": 0.9706,
"step": 33900
},
{
"epoch": 0.95,
"learning_rate": 4.3273623609509996e-06,
"loss": 0.9749,
"step": 34000
},
{
"epoch": 0.95,
"learning_rate": 4.321832257731189e-06,
"loss": 0.969,
"step": 34100
},
{
"epoch": 0.95,
"learning_rate": 4.316283079372959e-06,
"loss": 1.0083,
"step": 34200
},
{
"epoch": 0.95,
"learning_rate": 4.3107148839779e-06,
"loss": 0.9969,
"step": 34300
},
{
"epoch": 0.96,
"learning_rate": 4.305127729846711e-06,
"loss": 0.9863,
"step": 34400
},
{
"epoch": 0.96,
"learning_rate": 4.299521675478598e-06,
"loss": 0.9646,
"step": 34500
},
{
"epoch": 0.96,
"learning_rate": 4.293896779570656e-06,
"loss": 0.9704,
"step": 34600
},
{
"epoch": 0.96,
"learning_rate": 4.288253101017259e-06,
"loss": 0.949,
"step": 34700
},
{
"epoch": 0.97,
"learning_rate": 4.282590698909439e-06,
"loss": 0.9667,
"step": 34800
},
{
"epoch": 0.97,
"learning_rate": 4.276909632534269e-06,
"loss": 0.9565,
"step": 34900
},
{
"epoch": 0.97,
"learning_rate": 4.271209961374246e-06,
"loss": 0.9733,
"step": 35000
},
{
"epoch": 0.97,
"eval_bleu": 7.9496,
"eval_gen_len": 18.9815,
"eval_loss": 0.9227670431137085,
"eval_runtime": 961.6825,
"eval_samples_per_second": 2.076,
"eval_steps_per_second": 1.038,
"step": 35000
},
{
"epoch": 0.98,
"learning_rate": 4.26549174510666e-06,
"loss": 0.9857,
"step": 35100
},
{
"epoch": 0.98,
"learning_rate": 4.259755043602978e-06,
"loss": 0.9431,
"step": 35200
},
{
"epoch": 0.98,
"learning_rate": 4.253999916928211e-06,
"loss": 0.9731,
"step": 35300
},
{
"epoch": 0.98,
"learning_rate": 4.248226425340288e-06,
"loss": 0.9833,
"step": 35400
},
{
"epoch": 0.99,
"learning_rate": 4.242434629289421e-06,
"loss": 0.9582,
"step": 35500
},
{
"epoch": 0.99,
"learning_rate": 4.236624589417482e-06,
"loss": 1.002,
"step": 35600
},
{
"epoch": 0.99,
"learning_rate": 4.230796366557354e-06,
"loss": 0.9598,
"step": 35700
},
{
"epoch": 1.0,
"learning_rate": 4.224950021732307e-06,
"loss": 0.9576,
"step": 35800
},
{
"epoch": 1.0,
"learning_rate": 4.21908561615535e-06,
"loss": 1.0025,
"step": 35900
},
{
"epoch": 1.0,
"learning_rate": 4.213203211228596e-06,
"loss": 0.9746,
"step": 36000
},
{
"epoch": 1.0,
"learning_rate": 4.2073028685426146e-06,
"loss": 0.9419,
"step": 36100
},
{
"epoch": 1.01,
"learning_rate": 4.20138464987579e-06,
"loss": 0.9092,
"step": 36200
},
{
"epoch": 1.01,
"learning_rate": 4.195448617193676e-06,
"loss": 0.9345,
"step": 36300
},
{
"epoch": 1.01,
"learning_rate": 4.189494832648339e-06,
"loss": 0.9613,
"step": 36400
},
{
"epoch": 1.02,
"learning_rate": 4.183523358577716e-06,
"loss": 0.9596,
"step": 36500
},
{
"epoch": 1.02,
"learning_rate": 4.177534257504961e-06,
"loss": 0.9412,
"step": 36600
},
{
"epoch": 1.02,
"learning_rate": 4.171527592137783e-06,
"loss": 0.9418,
"step": 36700
},
{
"epoch": 1.02,
"learning_rate": 4.1655034253678e-06,
"loss": 0.9166,
"step": 36800
},
{
"epoch": 1.03,
"learning_rate": 4.15946182026987e-06,
"loss": 0.9483,
"step": 36900
},
{
"epoch": 1.03,
"learning_rate": 4.153402840101438e-06,
"loss": 0.9604,
"step": 37000
},
{
"epoch": 1.03,
"learning_rate": 4.1473265483018735e-06,
"loss": 0.928,
"step": 37100
},
{
"epoch": 1.03,
"learning_rate": 4.141233008491797e-06,
"loss": 0.953,
"step": 37200
},
{
"epoch": 1.04,
"learning_rate": 4.1351222844724305e-06,
"loss": 0.935,
"step": 37300
},
{
"epoch": 1.04,
"learning_rate": 4.128994440224912e-06,
"loss": 0.9269,
"step": 37400
},
{
"epoch": 1.04,
"learning_rate": 4.122849539909637e-06,
"loss": 0.9423,
"step": 37500
},
{
"epoch": 1.05,
"learning_rate": 4.1166876478655835e-06,
"loss": 0.9402,
"step": 37600
},
{
"epoch": 1.05,
"learning_rate": 4.110508828609638e-06,
"loss": 0.9472,
"step": 37700
},
{
"epoch": 1.05,
"learning_rate": 4.1043131468359155e-06,
"loss": 0.9457,
"step": 37800
},
{
"epoch": 1.05,
"learning_rate": 4.098100667415095e-06,
"loss": 0.8999,
"step": 37900
},
{
"epoch": 1.06,
"learning_rate": 4.091871455393725e-06,
"loss": 0.9388,
"step": 38000
},
{
"epoch": 1.06,
"learning_rate": 4.0856255759935515e-06,
"loss": 0.9416,
"step": 38100
},
{
"epoch": 1.06,
"learning_rate": 4.079363094610836e-06,
"loss": 0.98,
"step": 38200
},
{
"epoch": 1.07,
"learning_rate": 4.0730840768156625e-06,
"loss": 0.9424,
"step": 38300
},
{
"epoch": 1.07,
"learning_rate": 4.066788588351261e-06,
"loss": 0.9192,
"step": 38400
},
{
"epoch": 1.07,
"learning_rate": 4.0604766951333105e-06,
"loss": 0.9188,
"step": 38500
},
{
"epoch": 1.07,
"learning_rate": 4.054148463249257e-06,
"loss": 0.9636,
"step": 38600
},
{
"epoch": 1.08,
"learning_rate": 4.0478039589576146e-06,
"loss": 0.9716,
"step": 38700
},
{
"epoch": 1.08,
"learning_rate": 4.041443248687273e-06,
"loss": 0.955,
"step": 38800
},
{
"epoch": 1.08,
"learning_rate": 4.035066399036807e-06,
"loss": 0.9519,
"step": 38900
},
{
"epoch": 1.08,
"learning_rate": 4.028673476773774e-06,
"loss": 0.9069,
"step": 39000
},
{
"epoch": 1.09,
"learning_rate": 4.022264548834016e-06,
"loss": 0.9085,
"step": 39100
},
{
"epoch": 1.09,
"learning_rate": 4.015839682320959e-06,
"loss": 0.9313,
"step": 39200
},
{
"epoch": 1.09,
"learning_rate": 4.0093989445049135e-06,
"loss": 0.9648,
"step": 39300
},
{
"epoch": 1.1,
"learning_rate": 4.002942402822364e-06,
"loss": 0.9398,
"step": 39400
},
{
"epoch": 1.1,
"learning_rate": 3.9964701248752665e-06,
"loss": 0.9305,
"step": 39500
},
{
"epoch": 1.1,
"learning_rate": 3.989982178430345e-06,
"loss": 0.9475,
"step": 39600
},
{
"epoch": 1.1,
"learning_rate": 3.983478631418372e-06,
"loss": 0.9448,
"step": 39700
},
{
"epoch": 1.11,
"learning_rate": 3.976959551933464e-06,
"loss": 0.9754,
"step": 39800
},
{
"epoch": 1.11,
"learning_rate": 3.970425008232369e-06,
"loss": 0.9564,
"step": 39900
},
{
"epoch": 1.11,
"learning_rate": 3.96387506873375e-06,
"loss": 0.9035,
"step": 40000
},
{
"epoch": 1.11,
"eval_bleu": 7.689,
"eval_gen_len": 18.9795,
"eval_loss": 0.9161636233329773,
"eval_runtime": 963.6243,
"eval_samples_per_second": 2.071,
"eval_steps_per_second": 1.036,
"step": 40000
},
{
"epoch": 1.12,
"learning_rate": 3.957309802017466e-06,
"loss": 0.9898,
"step": 40100
},
{
"epoch": 1.12,
"learning_rate": 3.95072927682386e-06,
"loss": 0.8963,
"step": 40200
},
{
"epoch": 1.12,
"learning_rate": 3.944133562053033e-06,
"loss": 0.9179,
"step": 40300
},
{
"epoch": 1.12,
"learning_rate": 3.937522726764128e-06,
"loss": 0.9368,
"step": 40400
},
{
"epoch": 1.13,
"learning_rate": 3.930896840174603e-06,
"loss": 0.9281,
"step": 40500
},
{
"epoch": 1.13,
"learning_rate": 3.924255971659506e-06,
"loss": 0.9706,
"step": 40600
},
{
"epoch": 1.13,
"learning_rate": 3.9176001907507546e-06,
"loss": 0.9401,
"step": 40700
},
{
"epoch": 1.13,
"learning_rate": 3.910929567136401e-06,
"loss": 0.9185,
"step": 40800
},
{
"epoch": 1.14,
"learning_rate": 3.904244170659904e-06,
"loss": 0.9406,
"step": 40900
},
{
"epoch": 1.14,
"learning_rate": 3.8975440713194016e-06,
"loss": 0.923,
"step": 41000
},
{
"epoch": 1.14,
"learning_rate": 3.890829339266973e-06,
"loss": 0.9427,
"step": 41100
},
{
"epoch": 1.15,
"learning_rate": 3.884100044807907e-06,
"loss": 0.9471,
"step": 41200
},
{
"epoch": 1.15,
"learning_rate": 3.877356258399967e-06,
"loss": 0.9367,
"step": 41300
},
{
"epoch": 1.15,
"learning_rate": 3.870598050652648e-06,
"loss": 0.9063,
"step": 41400
},
{
"epoch": 1.15,
"learning_rate": 3.863825492326444e-06,
"loss": 0.9604,
"step": 41500
},
{
"epoch": 1.16,
"learning_rate": 3.857038654332102e-06,
"loss": 0.8888,
"step": 41600
},
{
"epoch": 1.16,
"learning_rate": 3.850237607729882e-06,
"loss": 0.9494,
"step": 41700
},
{
"epoch": 1.16,
"learning_rate": 3.8434224237288134e-06,
"loss": 0.8925,
"step": 41800
},
{
"epoch": 1.17,
"learning_rate": 3.836593173685946e-06,
"loss": 0.9473,
"step": 41900
},
{
"epoch": 1.17,
"learning_rate": 3.829749929105609e-06,
"loss": 0.9236,
"step": 42000
},
{
"epoch": 1.17,
"learning_rate": 3.822892761638656e-06,
"loss": 0.9083,
"step": 42100
},
{
"epoch": 1.17,
"learning_rate": 3.816021743081717e-06,
"loss": 0.9135,
"step": 42200
},
{
"epoch": 1.18,
"learning_rate": 3.8091369453764504e-06,
"loss": 0.9526,
"step": 42300
},
{
"epoch": 1.18,
"learning_rate": 3.8022384406087824e-06,
"loss": 0.9345,
"step": 42400
},
{
"epoch": 1.18,
"learning_rate": 3.79532630100816e-06,
"loss": 0.9176,
"step": 42500
},
{
"epoch": 1.18,
"learning_rate": 3.7884005989467866e-06,
"loss": 0.9467,
"step": 42600
},
{
"epoch": 1.19,
"learning_rate": 3.781461406938874e-06,
"loss": 0.9271,
"step": 42700
},
{
"epoch": 1.19,
"learning_rate": 3.774508797639874e-06,
"loss": 0.9588,
"step": 42800
},
{
"epoch": 1.19,
"learning_rate": 3.7675428438457234e-06,
"loss": 0.9201,
"step": 42900
},
{
"epoch": 1.2,
"learning_rate": 3.760563618492079e-06,
"loss": 0.9285,
"step": 43000
},
{
"epoch": 1.2,
"learning_rate": 3.7535711946535552e-06,
"loss": 0.9484,
"step": 43100
},
{
"epoch": 1.2,
"learning_rate": 3.746565645542958e-06,
"loss": 0.9253,
"step": 43200
},
{
"epoch": 1.2,
"learning_rate": 3.739547044510521e-06,
"loss": 0.9179,
"step": 43300
},
{
"epoch": 1.21,
"learning_rate": 3.7325154650431317e-06,
"loss": 0.9143,
"step": 43400
},
{
"epoch": 1.21,
"learning_rate": 3.725470980763569e-06,
"loss": 0.9382,
"step": 43500
},
{
"epoch": 1.21,
"learning_rate": 3.718413665429729e-06,
"loss": 0.9533,
"step": 43600
},
{
"epoch": 1.22,
"learning_rate": 3.711343592933851e-06,
"loss": 0.9214,
"step": 43700
},
{
"epoch": 1.22,
"learning_rate": 3.704260837301746e-06,
"loss": 0.8943,
"step": 43800
},
{
"epoch": 1.22,
"learning_rate": 3.6971654726920243e-06,
"loss": 0.9063,
"step": 43900
},
{
"epoch": 1.22,
"learning_rate": 3.690057573395311e-06,
"loss": 0.8965,
"step": 44000
},
{
"epoch": 1.23,
"learning_rate": 3.6829372138334763e-06,
"loss": 0.9241,
"step": 44100
},
{
"epoch": 1.23,
"learning_rate": 3.6758044685588547e-06,
"loss": 0.9644,
"step": 44200
},
{
"epoch": 1.23,
"learning_rate": 3.668659412253458e-06,
"loss": 0.9391,
"step": 44300
},
{
"epoch": 1.23,
"learning_rate": 3.661502119728203e-06,
"loss": 0.9273,
"step": 44400
},
{
"epoch": 1.24,
"learning_rate": 3.6543326659221213e-06,
"loss": 0.9354,
"step": 44500
},
{
"epoch": 1.24,
"learning_rate": 3.6471511259015764e-06,
"loss": 0.9162,
"step": 44600
},
{
"epoch": 1.24,
"learning_rate": 3.6399575748594796e-06,
"loss": 0.9281,
"step": 44700
},
{
"epoch": 1.25,
"learning_rate": 3.6327520881145002e-06,
"loss": 0.8999,
"step": 44800
},
{
"epoch": 1.25,
"learning_rate": 3.6255347411102777e-06,
"loss": 0.909,
"step": 44900
},
{
"epoch": 1.25,
"learning_rate": 3.6183056094146333e-06,
"loss": 0.9386,
"step": 45000
},
{
"epoch": 1.25,
"eval_bleu": 7.6781,
"eval_gen_len": 18.9825,
"eval_loss": 0.9038894176483154,
"eval_runtime": 963.1893,
"eval_samples_per_second": 2.072,
"eval_steps_per_second": 1.036,
"step": 45000
},
{
"epoch": 1.25,
"learning_rate": 3.611064768718777e-06,
"loss": 0.8965,
"step": 45100
},
{
"epoch": 1.26,
"learning_rate": 3.603812294836515e-06,
"loss": 0.9717,
"step": 45200
},
{
"epoch": 1.26,
"learning_rate": 3.5965482637034567e-06,
"loss": 0.8826,
"step": 45300
},
{
"epoch": 1.26,
"learning_rate": 3.58927275137622e-06,
"loss": 0.9187,
"step": 45400
},
{
"epoch": 1.27,
"learning_rate": 3.581985834031635e-06,
"loss": 0.888,
"step": 45500
},
{
"epoch": 1.27,
"learning_rate": 3.5746875879659426e-06,
"loss": 0.9339,
"step": 45600
},
{
"epoch": 1.27,
"learning_rate": 3.5673780895940034e-06,
"loss": 0.9291,
"step": 45700
},
{
"epoch": 1.27,
"learning_rate": 3.56005741544849e-06,
"loss": 0.9303,
"step": 45800
},
{
"epoch": 1.28,
"learning_rate": 3.5527256421790902e-06,
"loss": 0.9123,
"step": 45900
},
{
"epoch": 1.28,
"learning_rate": 3.5453828465517e-06,
"loss": 0.9037,
"step": 46000
},
{
"epoch": 1.28,
"learning_rate": 3.538029105447628e-06,
"loss": 0.9447,
"step": 46100
},
{
"epoch": 1.28,
"learning_rate": 3.530664495862782e-06,
"loss": 0.9011,
"step": 46200
},
{
"epoch": 1.29,
"learning_rate": 3.523289094906865e-06,
"loss": 0.8897,
"step": 46300
},
{
"epoch": 1.29,
"learning_rate": 3.5159029798025717e-06,
"loss": 0.9278,
"step": 46400
},
{
"epoch": 1.29,
"learning_rate": 3.5085062278847765e-06,
"loss": 0.9204,
"step": 46500
},
{
"epoch": 1.3,
"learning_rate": 3.5010989165997227e-06,
"loss": 0.9185,
"step": 46600
},
{
"epoch": 1.3,
"learning_rate": 3.4936811235042158e-06,
"loss": 0.9067,
"step": 46700
},
{
"epoch": 1.3,
"learning_rate": 3.4862529262648076e-06,
"loss": 0.9469,
"step": 46800
},
{
"epoch": 1.3,
"learning_rate": 3.4788144026569846e-06,
"loss": 0.942,
"step": 46900
},
{
"epoch": 1.31,
"learning_rate": 3.4713656305643543e-06,
"loss": 0.9493,
"step": 47000
},
{
"epoch": 1.31,
"learning_rate": 3.4639066879778278e-06,
"loss": 0.9285,
"step": 47100
},
{
"epoch": 1.31,
"learning_rate": 3.4564376529948045e-06,
"loss": 0.9242,
"step": 47200
},
{
"epoch": 1.32,
"learning_rate": 3.4489586038183564e-06,
"loss": 0.9045,
"step": 47300
},
{
"epoch": 1.32,
"learning_rate": 3.4414696187564035e-06,
"loss": 0.9233,
"step": 47400
},
{
"epoch": 1.32,
"learning_rate": 3.4339707762209006e-06,
"loss": 0.896,
"step": 47500
},
{
"epoch": 1.32,
"learning_rate": 3.426462154727012e-06,
"loss": 0.9072,
"step": 47600
},
{
"epoch": 1.33,
"learning_rate": 3.418943832892291e-06,
"loss": 0.9419,
"step": 47700
},
{
"epoch": 1.33,
"learning_rate": 3.411415889435856e-06,
"loss": 0.8977,
"step": 47800
},
{
"epoch": 1.33,
"learning_rate": 3.403878403177567e-06,
"loss": 0.8837,
"step": 47900
},
{
"epoch": 1.33,
"learning_rate": 3.3963314530372e-06,
"loss": 0.9148,
"step": 48000
},
{
"epoch": 1.34,
"learning_rate": 3.388775118033621e-06,
"loss": 0.9005,
"step": 48100
},
{
"epoch": 1.34,
"learning_rate": 3.381209477283957e-06,
"loss": 0.8907,
"step": 48200
},
{
"epoch": 1.34,
"learning_rate": 3.3736346100027717e-06,
"loss": 0.9064,
"step": 48300
},
{
"epoch": 1.35,
"learning_rate": 3.3660505955012308e-06,
"loss": 0.9026,
"step": 48400
},
{
"epoch": 1.35,
"learning_rate": 3.3584575131862757e-06,
"loss": 0.9501,
"step": 48500
},
{
"epoch": 1.35,
"learning_rate": 3.3508554425597896e-06,
"loss": 0.9016,
"step": 48600
},
{
"epoch": 1.35,
"learning_rate": 3.3432444632177663e-06,
"loss": 0.9159,
"step": 48700
},
{
"epoch": 1.36,
"learning_rate": 3.335624654849477e-06,
"loss": 0.8766,
"step": 48800
},
{
"epoch": 1.36,
"learning_rate": 3.327996097236636e-06,
"loss": 0.9125,
"step": 48900
},
{
"epoch": 1.36,
"learning_rate": 3.320358870252563e-06,
"loss": 0.8931,
"step": 49000
},
{
"epoch": 1.37,
"learning_rate": 3.3127130538613506e-06,
"loss": 0.9088,
"step": 49100
},
{
"epoch": 1.37,
"learning_rate": 3.3050587281170245e-06,
"loss": 0.9427,
"step": 49200
},
{
"epoch": 1.37,
"learning_rate": 3.297395973162705e-06,
"loss": 0.9042,
"step": 49300
},
{
"epoch": 1.37,
"learning_rate": 3.2897248692297678e-06,
"loss": 0.9094,
"step": 49400
},
{
"epoch": 1.38,
"learning_rate": 3.2820454966370102e-06,
"loss": 0.9384,
"step": 49500
},
{
"epoch": 1.38,
"learning_rate": 3.2743579357897997e-06,
"loss": 0.888,
"step": 49600
},
{
"epoch": 1.38,
"learning_rate": 3.266662267179238e-06,
"loss": 0.9208,
"step": 49700
},
{
"epoch": 1.38,
"learning_rate": 3.25895857138132e-06,
"loss": 0.9304,
"step": 49800
},
{
"epoch": 1.39,
"learning_rate": 3.2512469290560848e-06,
"loss": 0.9346,
"step": 49900
},
{
"epoch": 1.39,
"learning_rate": 3.2435274209467765e-06,
"loss": 0.9073,
"step": 50000
},
{
"epoch": 1.39,
"eval_bleu": 7.8607,
"eval_gen_len": 18.9805,
"eval_loss": 0.8985511064529419,
"eval_runtime": 966.2977,
"eval_samples_per_second": 2.066,
"eval_steps_per_second": 1.033,
"step": 50000
},
{
"epoch": 1.39,
"learning_rate": 3.235800127878995e-06,
"loss": 0.8804,
"step": 50100
},
{
"epoch": 1.4,
"learning_rate": 3.228065130759852e-06,
"loss": 0.9301,
"step": 50200
},
{
"epoch": 1.4,
"learning_rate": 3.220322510577121e-06,
"loss": 0.9109,
"step": 50300
},
{
"epoch": 1.4,
"learning_rate": 3.2125723483983935e-06,
"loss": 0.9029,
"step": 50400
},
{
"epoch": 1.4,
"learning_rate": 3.204814725370227e-06,
"loss": 0.9089,
"step": 50500
},
{
"epoch": 1.41,
"learning_rate": 3.1970497227172957e-06,
"loss": 0.9184,
"step": 50600
},
{
"epoch": 1.41,
"learning_rate": 3.1892774217415433e-06,
"loss": 0.894,
"step": 50700
},
{
"epoch": 1.41,
"learning_rate": 3.181497903821326e-06,
"loss": 0.905,
"step": 50800
},
{
"epoch": 1.42,
"learning_rate": 3.1737112504105655e-06,
"loss": 0.889,
"step": 50900
},
{
"epoch": 1.42,
"learning_rate": 3.165917543037894e-06,
"loss": 0.8977,
"step": 51000
},
{
"epoch": 1.42,
"learning_rate": 3.1581168633058002e-06,
"loss": 0.9111,
"step": 51100
},
{
"epoch": 1.42,
"learning_rate": 3.150309292889776e-06,
"loss": 0.9073,
"step": 51200
},
{
"epoch": 1.43,
"learning_rate": 3.14249491353746e-06,
"loss": 0.9229,
"step": 51300
},
{
"epoch": 1.43,
"learning_rate": 3.134673807067784e-06,
"loss": 0.8961,
"step": 51400
},
{
"epoch": 1.43,
"learning_rate": 3.1268460553701146e-06,
"loss": 0.8976,
"step": 51500
},
{
"epoch": 1.43,
"learning_rate": 3.1190117404033943e-06,
"loss": 0.9186,
"step": 51600
},
{
"epoch": 1.44,
"learning_rate": 3.111170944195286e-06,
"loss": 0.8826,
"step": 51700
},
{
"epoch": 1.44,
"learning_rate": 3.1033237488413144e-06,
"loss": 0.9097,
"step": 51800
},
{
"epoch": 1.44,
"learning_rate": 3.095470236504003e-06,
"loss": 0.8805,
"step": 51900
},
{
"epoch": 1.45,
"learning_rate": 3.0876104894120164e-06,
"loss": 0.8718,
"step": 52000
},
{
"epoch": 1.45,
"learning_rate": 3.0797445898593007e-06,
"loss": 0.9029,
"step": 52100
},
{
"epoch": 1.45,
"learning_rate": 3.07187262020422e-06,
"loss": 0.9297,
"step": 52200
},
{
"epoch": 1.45,
"learning_rate": 3.0639946628686913e-06,
"loss": 0.8916,
"step": 52300
},
{
"epoch": 1.46,
"learning_rate": 3.0561108003373275e-06,
"loss": 0.9609,
"step": 52400
},
{
"epoch": 1.46,
"learning_rate": 3.0482211151565693e-06,
"loss": 0.8926,
"step": 52500
},
{
"epoch": 1.46,
"learning_rate": 3.0403256899338236e-06,
"loss": 0.8811,
"step": 52600
},
{
"epoch": 1.47,
"learning_rate": 3.032424607336595e-06,
"loss": 0.8748,
"step": 52700
},
{
"epoch": 1.47,
"learning_rate": 3.0245179500916245e-06,
"loss": 0.9036,
"step": 52800
},
{
"epoch": 1.47,
"learning_rate": 3.016605800984021e-06,
"loss": 0.9,
"step": 52900
},
{
"epoch": 1.47,
"learning_rate": 3.0086882428563948e-06,
"loss": 0.8919,
"step": 53000
},
{
"epoch": 1.48,
"learning_rate": 3.0007653586079884e-06,
"loss": 0.924,
"step": 53100
},
{
"epoch": 1.48,
"learning_rate": 2.9928372311938134e-06,
"loss": 0.9289,
"step": 53200
},
{
"epoch": 1.48,
"learning_rate": 2.984903943623779e-06,
"loss": 0.8922,
"step": 53300
},
{
"epoch": 1.49,
"learning_rate": 2.9769655789618185e-06,
"loss": 0.904,
"step": 53400
},
{
"epoch": 1.49,
"learning_rate": 2.9690222203250286e-06,
"loss": 0.9239,
"step": 53500
},
{
"epoch": 1.49,
"learning_rate": 2.961073950882793e-06,
"loss": 0.9248,
"step": 53600
},
{
"epoch": 1.49,
"learning_rate": 2.9531208538559114e-06,
"loss": 0.8867,
"step": 53700
},
{
"epoch": 1.5,
"learning_rate": 2.945163012515732e-06,
"loss": 0.8872,
"step": 53800
},
{
"epoch": 1.5,
"learning_rate": 2.9372005101832767e-06,
"loss": 0.9082,
"step": 53900
},
{
"epoch": 1.5,
"learning_rate": 2.9292334302283683e-06,
"loss": 0.8745,
"step": 54000
},
{
"epoch": 1.5,
"learning_rate": 2.9212618560687604e-06,
"loss": 0.8917,
"step": 54100
},
{
"epoch": 1.51,
"learning_rate": 2.9132858711692607e-06,
"loss": 0.8771,
"step": 54200
},
{
"epoch": 1.51,
"learning_rate": 2.9053055590408603e-06,
"loss": 0.8999,
"step": 54300
},
{
"epoch": 1.51,
"learning_rate": 2.8973210032398567e-06,
"loss": 0.8902,
"step": 54400
},
{
"epoch": 1.52,
"learning_rate": 2.88933228736698e-06,
"loss": 0.8912,
"step": 54500
},
{
"epoch": 1.52,
"learning_rate": 2.881339495066518e-06,
"loss": 0.8482,
"step": 54600
},
{
"epoch": 1.52,
"learning_rate": 2.8733427100254383e-06,
"loss": 0.9028,
"step": 54700
},
{
"epoch": 1.52,
"learning_rate": 2.8653420159725166e-06,
"loss": 0.8881,
"step": 54800
},
{
"epoch": 1.53,
"learning_rate": 2.8573374966774546e-06,
"loss": 0.8913,
"step": 54900
},
{
"epoch": 1.53,
"learning_rate": 2.849329235950007e-06,
"loss": 0.8928,
"step": 55000
},
{
"epoch": 1.53,
"eval_bleu": 8.0666,
"eval_gen_len": 18.981,
"eval_loss": 0.8941593170166016,
"eval_runtime": 968.3576,
"eval_samples_per_second": 2.061,
"eval_steps_per_second": 1.031,
"step": 55000
},
{
"epoch": 1.53,
"learning_rate": 2.8413173176391006e-06,
"loss": 0.8992,
"step": 55100
},
{
"epoch": 1.54,
"learning_rate": 2.8333018256319617e-06,
"loss": 0.8777,
"step": 55200
},
{
"epoch": 1.54,
"learning_rate": 2.82528284385323e-06,
"loss": 0.8709,
"step": 55300
},
{
"epoch": 1.54,
"learning_rate": 2.817260456264086e-06,
"loss": 0.9326,
"step": 55400
},
{
"epoch": 1.54,
"learning_rate": 2.809234746861372e-06,
"loss": 0.9006,
"step": 55500
},
{
"epoch": 1.55,
"learning_rate": 2.801205799676709e-06,
"loss": 0.929,
"step": 55600
},
{
"epoch": 1.55,
"learning_rate": 2.7931736987756165e-06,
"loss": 0.9161,
"step": 55700
},
{
"epoch": 1.55,
"learning_rate": 2.7851385282566372e-06,
"loss": 0.908,
"step": 55800
},
{
"epoch": 1.55,
"learning_rate": 2.7771003722504534e-06,
"loss": 0.9008,
"step": 55900
},
{
"epoch": 1.56,
"learning_rate": 2.769059314919006e-06,
"loss": 0.9195,
"step": 56000
},
{
"epoch": 1.56,
"learning_rate": 2.7610154404546136e-06,
"loss": 0.8826,
"step": 56100
},
{
"epoch": 1.56,
"learning_rate": 2.752968833079089e-06,
"loss": 0.858,
"step": 56200
},
{
"epoch": 1.57,
"learning_rate": 2.744919577042863e-06,
"loss": 0.9305,
"step": 56300
},
{
"epoch": 1.57,
"learning_rate": 2.7368677566240976e-06,
"loss": 0.9289,
"step": 56400
},
{
"epoch": 1.57,
"learning_rate": 2.7288134561278017e-06,
"loss": 0.9168,
"step": 56500
},
{
"epoch": 1.57,
"learning_rate": 2.720756759884956e-06,
"loss": 0.8958,
"step": 56600
},
{
"epoch": 1.58,
"learning_rate": 2.7126977522516223e-06,
"loss": 0.8924,
"step": 56700
},
{
"epoch": 1.58,
"learning_rate": 2.7046365176080635e-06,
"loss": 0.9235,
"step": 56800
},
{
"epoch": 1.58,
"learning_rate": 2.6965731403578614e-06,
"loss": 0.9171,
"step": 56900
},
{
"epoch": 1.59,
"learning_rate": 2.6885077049270316e-06,
"loss": 0.9159,
"step": 57000
},
{
"epoch": 1.59,
"learning_rate": 2.6804402957631364e-06,
"loss": 0.8949,
"step": 57100
},
{
"epoch": 1.59,
"learning_rate": 2.6723709973344088e-06,
"loss": 0.8796,
"step": 57200
},
{
"epoch": 1.59,
"learning_rate": 2.6642998941288573e-06,
"loss": 0.8943,
"step": 57300
},
{
"epoch": 1.6,
"learning_rate": 2.6562270706533917e-06,
"loss": 0.8714,
"step": 57400
},
{
"epoch": 1.6,
"learning_rate": 2.6481526114329313e-06,
"loss": 0.8647,
"step": 57500
},
{
"epoch": 1.6,
"learning_rate": 2.640076601009522e-06,
"loss": 0.9151,
"step": 57600
},
{
"epoch": 1.6,
"learning_rate": 2.631999123941452e-06,
"loss": 0.8942,
"step": 57700
},
{
"epoch": 1.61,
"learning_rate": 2.6239202648023666e-06,
"loss": 0.8863,
"step": 57800
},
{
"epoch": 1.61,
"learning_rate": 2.6158401081803784e-06,
"loss": 0.9138,
"step": 57900
},
{
"epoch": 1.61,
"learning_rate": 2.6077587386771896e-06,
"loss": 0.9095,
"step": 58000
},
{
"epoch": 1.62,
"learning_rate": 2.5996762409071978e-06,
"loss": 0.9049,
"step": 58100
},
{
"epoch": 1.62,
"learning_rate": 2.591592699496616e-06,
"loss": 0.8793,
"step": 58200
},
{
"epoch": 1.62,
"learning_rate": 2.583508199082585e-06,
"loss": 0.864,
"step": 58300
},
{
"epoch": 1.62,
"learning_rate": 2.575422824312284e-06,
"loss": 0.9173,
"step": 58400
},
{
"epoch": 1.63,
"learning_rate": 2.5673366598420487e-06,
"loss": 0.8799,
"step": 58500
},
{
"epoch": 1.63,
"learning_rate": 2.5592497903364834e-06,
"loss": 0.8751,
"step": 58600
},
{
"epoch": 1.63,
"learning_rate": 2.5511623004675743e-06,
"loss": 0.8871,
"step": 58700
},
{
"epoch": 1.64,
"learning_rate": 2.5430742749138015e-06,
"loss": 0.9121,
"step": 58800
},
{
"epoch": 1.64,
"learning_rate": 2.534985798359257e-06,
"loss": 0.8947,
"step": 58900
},
{
"epoch": 1.64,
"learning_rate": 2.5268969554927512e-06,
"loss": 0.8911,
"step": 59000
},
{
"epoch": 1.64,
"learning_rate": 2.5188078310069326e-06,
"loss": 0.8626,
"step": 59100
},
{
"epoch": 1.65,
"learning_rate": 2.5107185095973967e-06,
"loss": 0.8943,
"step": 59200
},
{
"epoch": 1.65,
"learning_rate": 2.5026290759618026e-06,
"loss": 0.9163,
"step": 59300
},
{
"epoch": 1.65,
"learning_rate": 2.494539614798982e-06,
"loss": 0.8599,
"step": 59400
},
{
"epoch": 1.65,
"learning_rate": 2.486450210808057e-06,
"loss": 0.8907,
"step": 59500
},
{
"epoch": 1.66,
"learning_rate": 2.4783609486875507e-06,
"loss": 0.889,
"step": 59600
},
{
"epoch": 1.66,
"learning_rate": 2.4702719131345003e-06,
"loss": 0.8872,
"step": 59700
},
{
"epoch": 1.66,
"learning_rate": 2.462183188843569e-06,
"loss": 0.8836,
"step": 59800
},
{
"epoch": 1.67,
"learning_rate": 2.4540948605061652e-06,
"loss": 0.8837,
"step": 59900
},
{
"epoch": 1.67,
"learning_rate": 2.446007012809548e-06,
"loss": 0.884,
"step": 60000
},
{
"epoch": 1.67,
"eval_bleu": 8.1679,
"eval_gen_len": 18.9785,
"eval_loss": 0.8873680830001831,
"eval_runtime": 968.5261,
"eval_samples_per_second": 2.061,
"eval_steps_per_second": 1.03,
"step": 60000
},
{
"epoch": 1.67,
"learning_rate": 2.437919730435946e-06,
"loss": 0.8902,
"step": 60100
},
{
"epoch": 1.67,
"learning_rate": 2.4298330980616674e-06,
"loss": 0.8499,
"step": 60200
},
{
"epoch": 1.68,
"learning_rate": 2.4217472003562144e-06,
"loss": 0.8927,
"step": 60300
},
{
"epoch": 1.68,
"learning_rate": 2.413662121981399e-06,
"loss": 0.9325,
"step": 60400
},
{
"epoch": 1.68,
"learning_rate": 2.4055779475904536e-06,
"loss": 0.8586,
"step": 60500
},
{
"epoch": 1.69,
"learning_rate": 2.397494761827145e-06,
"loss": 0.8788,
"step": 60600
},
{
"epoch": 1.69,
"learning_rate": 2.3894126493248884e-06,
"loss": 0.9161,
"step": 60700
},
{
"epoch": 1.69,
"learning_rate": 2.3813316947058634e-06,
"loss": 0.9058,
"step": 60800
},
{
"epoch": 1.69,
"learning_rate": 2.373251982580124e-06,
"loss": 0.8881,
"step": 60900
},
{
"epoch": 1.7,
"learning_rate": 2.365173597544718e-06,
"loss": 0.9027,
"step": 61000
},
{
"epoch": 1.7,
"learning_rate": 2.3570966241827947e-06,
"loss": 0.914,
"step": 61100
},
{
"epoch": 1.7,
"learning_rate": 2.3490211470627254e-06,
"loss": 0.8989,
"step": 61200
},
{
"epoch": 1.7,
"learning_rate": 2.3409472507372134e-06,
"loss": 0.8792,
"step": 61300
},
{
"epoch": 1.71,
"learning_rate": 2.3328750197424115e-06,
"loss": 0.8651,
"step": 61400
},
{
"epoch": 1.71,
"learning_rate": 2.3248045385970357e-06,
"loss": 0.9071,
"step": 61500
},
{
"epoch": 1.71,
"learning_rate": 2.316735891801482e-06,
"loss": 0.8676,
"step": 61600
},
{
"epoch": 1.72,
"learning_rate": 2.308669163836938e-06,
"loss": 0.9145,
"step": 61700
},
{
"epoch": 1.72,
"learning_rate": 2.300604439164501e-06,
"loss": 0.9072,
"step": 61800
},
{
"epoch": 1.72,
"learning_rate": 2.2925418022242955e-06,
"loss": 0.8916,
"step": 61900
},
{
"epoch": 1.72,
"learning_rate": 2.2844813374345837e-06,
"loss": 0.872,
"step": 62000
},
{
"epoch": 1.73,
"learning_rate": 2.2764231291908847e-06,
"loss": 0.8983,
"step": 62100
},
{
"epoch": 1.73,
"learning_rate": 2.2683672618650945e-06,
"loss": 0.8939,
"step": 62200
},
{
"epoch": 1.73,
"learning_rate": 2.2603138198045966e-06,
"loss": 0.9226,
"step": 62300
},
{
"epoch": 1.74,
"learning_rate": 2.2522628873313806e-06,
"loss": 0.8937,
"step": 62400
},
{
"epoch": 1.74,
"learning_rate": 2.2442145487411605e-06,
"loss": 0.8331,
"step": 62500
},
{
"epoch": 1.74,
"learning_rate": 2.2361688883024912e-06,
"loss": 0.8817,
"step": 62600
},
{
"epoch": 1.74,
"learning_rate": 2.228125990255889e-06,
"loss": 0.875,
"step": 62700
},
{
"epoch": 1.75,
"learning_rate": 2.2200859388129447e-06,
"loss": 0.9098,
"step": 62800
},
{
"epoch": 1.75,
"learning_rate": 2.2120488181554433e-06,
"loss": 0.8639,
"step": 62900
},
{
"epoch": 1.75,
"learning_rate": 2.2040147124344864e-06,
"loss": 0.89,
"step": 63000
},
{
"epoch": 1.75,
"learning_rate": 2.195983705769607e-06,
"loss": 0.8963,
"step": 63100
},
{
"epoch": 1.76,
"learning_rate": 2.1879558822478883e-06,
"loss": 0.9029,
"step": 63200
},
{
"epoch": 1.76,
"learning_rate": 2.1799313259230894e-06,
"loss": 0.8764,
"step": 63300
},
{
"epoch": 1.76,
"learning_rate": 2.1719101208147557e-06,
"loss": 0.8964,
"step": 63400
},
{
"epoch": 1.77,
"learning_rate": 2.163892350907349e-06,
"loss": 0.9076,
"step": 63500
},
{
"epoch": 1.77,
"learning_rate": 2.1558781001493604e-06,
"loss": 0.8936,
"step": 63600
},
{
"epoch": 1.77,
"learning_rate": 2.147867452452435e-06,
"loss": 0.8659,
"step": 63700
},
{
"epoch": 1.77,
"learning_rate": 2.1398604916904923e-06,
"loss": 0.897,
"step": 63800
},
{
"epoch": 1.78,
"learning_rate": 2.131857301698852e-06,
"loss": 0.8884,
"step": 63900
},
{
"epoch": 1.78,
"learning_rate": 2.123857966273348e-06,
"loss": 0.8893,
"step": 64000
},
{
"epoch": 1.78,
"learning_rate": 2.115862569169458e-06,
"loss": 0.9308,
"step": 64100
},
{
"epoch": 1.79,
"learning_rate": 2.1078711941014242e-06,
"loss": 0.8899,
"step": 64200
},
{
"epoch": 1.79,
"learning_rate": 2.099883924741376e-06,
"loss": 0.8964,
"step": 64300
},
{
"epoch": 1.79,
"learning_rate": 2.0919008447184562e-06,
"loss": 0.9132,
"step": 64400
},
{
"epoch": 1.79,
"learning_rate": 2.083922037617943e-06,
"loss": 0.9075,
"step": 64500
},
{
"epoch": 1.8,
"learning_rate": 2.075947586980376e-06,
"loss": 0.8968,
"step": 64600
},
{
"epoch": 1.8,
"learning_rate": 2.067977576300682e-06,
"loss": 0.8683,
"step": 64700
},
{
"epoch": 1.8,
"learning_rate": 2.0600120890272976e-06,
"loss": 0.8596,
"step": 64800
},
{
"epoch": 1.8,
"learning_rate": 2.0520512085613e-06,
"loss": 0.8863,
"step": 64900
},
{
"epoch": 1.81,
"learning_rate": 2.0440950182555337e-06,
"loss": 0.8786,
"step": 65000
},
{
"epoch": 1.81,
"eval_bleu": 7.8516,
"eval_gen_len": 18.9805,
"eval_loss": 0.8830544948577881,
"eval_runtime": 968.5663,
"eval_samples_per_second": 2.061,
"eval_steps_per_second": 1.03,
"step": 65000
},
{
"epoch": 1.81,
"learning_rate": 2.0361436014137315e-06,
"loss": 0.8911,
"step": 65100
},
{
"epoch": 1.81,
"learning_rate": 2.028197041289649e-06,
"loss": 0.9172,
"step": 65200
},
{
"epoch": 1.82,
"learning_rate": 2.0202554210861906e-06,
"loss": 0.8468,
"step": 65300
},
{
"epoch": 1.82,
"learning_rate": 2.0123188239545375e-06,
"loss": 0.8445,
"step": 65400
},
{
"epoch": 1.82,
"learning_rate": 2.0043873329932774e-06,
"loss": 0.8917,
"step": 65500
},
{
"epoch": 1.82,
"learning_rate": 1.996461031247536e-06,
"loss": 0.8759,
"step": 65600
},
{
"epoch": 1.83,
"learning_rate": 1.9885400017081075e-06,
"loss": 0.9045,
"step": 65700
},
{
"epoch": 1.83,
"learning_rate": 1.9806243273105807e-06,
"loss": 0.8706,
"step": 65800
},
{
"epoch": 1.83,
"learning_rate": 1.9727140909344767e-06,
"loss": 0.8973,
"step": 65900
},
{
"epoch": 1.84,
"learning_rate": 1.9648093754023784e-06,
"loss": 0.8648,
"step": 66000
},
{
"epoch": 1.84,
"learning_rate": 1.956910263479066e-06,
"loss": 0.8774,
"step": 66100
},
{
"epoch": 1.84,
"learning_rate": 1.9490168378706456e-06,
"loss": 0.9013,
"step": 66200
},
{
"epoch": 1.84,
"learning_rate": 1.9411291812236855e-06,
"loss": 0.9092,
"step": 66300
},
{
"epoch": 1.85,
"learning_rate": 1.9332473761243532e-06,
"loss": 0.8628,
"step": 66400
},
{
"epoch": 1.85,
"learning_rate": 1.925371505097548e-06,
"loss": 0.8867,
"step": 66500
},
{
"epoch": 1.85,
"learning_rate": 1.9175016506060357e-06,
"loss": 0.8694,
"step": 66600
},
{
"epoch": 1.85,
"learning_rate": 1.9096378950495915e-06,
"loss": 0.8613,
"step": 66700
},
{
"epoch": 1.86,
"learning_rate": 1.9017803207641282e-06,
"loss": 0.8966,
"step": 66800
},
{
"epoch": 1.86,
"learning_rate": 1.8939290100208425e-06,
"loss": 0.8381,
"step": 66900
},
{
"epoch": 1.86,
"learning_rate": 1.8860840450253467e-06,
"loss": 0.866,
"step": 67000
},
{
"epoch": 1.87,
"learning_rate": 1.8782455079168144e-06,
"loss": 0.9077,
"step": 67100
},
{
"epoch": 1.87,
"learning_rate": 1.8704134807671138e-06,
"loss": 0.908,
"step": 67200
},
{
"epoch": 1.87,
"learning_rate": 1.8625880455799562e-06,
"loss": 0.8655,
"step": 67300
},
{
"epoch": 1.87,
"learning_rate": 1.8547692842900283e-06,
"loss": 0.8985,
"step": 67400
},
{
"epoch": 1.88,
"learning_rate": 1.8469572787621426e-06,
"loss": 0.8852,
"step": 67500
},
{
"epoch": 1.88,
"learning_rate": 1.8391521107903747e-06,
"loss": 0.8918,
"step": 67600
},
{
"epoch": 1.88,
"learning_rate": 1.8313538620972094e-06,
"loss": 0.8765,
"step": 67700
},
{
"epoch": 1.89,
"learning_rate": 1.8235626143326865e-06,
"loss": 0.8762,
"step": 67800
},
{
"epoch": 1.89,
"learning_rate": 1.8157784490735404e-06,
"loss": 0.9053,
"step": 67900
},
{
"epoch": 1.89,
"learning_rate": 1.8080014478223523e-06,
"loss": 0.8811,
"step": 68000
},
{
"epoch": 1.89,
"learning_rate": 1.8002316920066932e-06,
"loss": 0.8794,
"step": 68100
},
{
"epoch": 1.9,
"learning_rate": 1.792469262978271e-06,
"loss": 0.8885,
"step": 68200
},
{
"epoch": 1.9,
"learning_rate": 1.7847142420120815e-06,
"loss": 0.8312,
"step": 68300
},
{
"epoch": 1.9,
"learning_rate": 1.7769667103055564e-06,
"loss": 0.856,
"step": 68400
},
{
"epoch": 1.9,
"learning_rate": 1.7692267489777104e-06,
"loss": 0.8706,
"step": 68500
},
{
"epoch": 1.91,
"learning_rate": 1.761494439068295e-06,
"loss": 0.873,
"step": 68600
},
{
"epoch": 1.91,
"learning_rate": 1.7537698615369504e-06,
"loss": 0.8709,
"step": 68700
},
{
"epoch": 1.91,
"learning_rate": 1.7460530972623537e-06,
"loss": 0.8237,
"step": 68800
},
{
"epoch": 1.92,
"learning_rate": 1.738344227041376e-06,
"loss": 0.8571,
"step": 68900
},
{
"epoch": 1.92,
"learning_rate": 1.7306433315882372e-06,
"loss": 0.8635,
"step": 69000
},
{
"epoch": 1.92,
"learning_rate": 1.7229504915336574e-06,
"loss": 0.8814,
"step": 69100
},
{
"epoch": 1.92,
"learning_rate": 1.715265787424013e-06,
"loss": 0.8667,
"step": 69200
},
{
"epoch": 1.93,
"learning_rate": 1.7075892997204958e-06,
"loss": 0.8739,
"step": 69300
},
{
"epoch": 1.93,
"learning_rate": 1.6999211087982686e-06,
"loss": 0.8479,
"step": 69400
},
{
"epoch": 1.93,
"learning_rate": 1.6922612949456274e-06,
"loss": 0.8878,
"step": 69500
},
{
"epoch": 1.94,
"learning_rate": 1.6846099383631537e-06,
"loss": 0.8956,
"step": 69600
},
{
"epoch": 1.94,
"learning_rate": 1.6769671191628807e-06,
"loss": 0.8428,
"step": 69700
},
{
"epoch": 1.94,
"learning_rate": 1.6693329173674521e-06,
"loss": 0.8344,
"step": 69800
},
{
"epoch": 1.94,
"learning_rate": 1.6617074129092857e-06,
"loss": 0.8504,
"step": 69900
},
{
"epoch": 1.95,
"learning_rate": 1.6540906856297336e-06,
"loss": 0.8899,
"step": 70000
},
{
"epoch": 1.95,
"eval_bleu": 7.9392,
"eval_gen_len": 18.9785,
"eval_loss": 0.8788951635360718,
"eval_runtime": 967.0825,
"eval_samples_per_second": 2.064,
"eval_steps_per_second": 1.032,
"step": 70000
},
{
"epoch": 1.95,
"learning_rate": 1.6464828152782508e-06,
"loss": 0.8444,
"step": 70100
},
{
"epoch": 1.95,
"learning_rate": 1.638883881511556e-06,
"loss": 0.8604,
"step": 70200
},
{
"epoch": 1.96,
"learning_rate": 1.6312939638927994e-06,
"loss": 0.8953,
"step": 70300
},
{
"epoch": 1.96,
"learning_rate": 1.623713141890728e-06,
"loss": 0.8298,
"step": 70400
},
{
"epoch": 1.96,
"learning_rate": 1.6161414948788575e-06,
"loss": 0.8403,
"step": 70500
},
{
"epoch": 1.96,
"learning_rate": 1.6085791021346365e-06,
"loss": 0.8284,
"step": 70600
},
{
"epoch": 1.97,
"learning_rate": 1.6010260428386205e-06,
"loss": 0.8546,
"step": 70700
},
{
"epoch": 1.97,
"learning_rate": 1.5934823960736402e-06,
"loss": 0.8742,
"step": 70800
},
{
"epoch": 1.97,
"learning_rate": 1.5859482408239718e-06,
"loss": 0.8856,
"step": 70900
},
{
"epoch": 1.97,
"learning_rate": 1.5784236559745175e-06,
"loss": 0.8231,
"step": 71000
},
{
"epoch": 1.98,
"learning_rate": 1.5709087203099687e-06,
"loss": 0.8579,
"step": 71100
},
{
"epoch": 1.98,
"learning_rate": 1.5634035125139923e-06,
"loss": 0.8614,
"step": 71200
},
{
"epoch": 1.98,
"learning_rate": 1.5559081111683977e-06,
"loss": 0.8646,
"step": 71300
},
{
"epoch": 1.99,
"learning_rate": 1.5484225947523201e-06,
"loss": 0.8503,
"step": 71400
},
{
"epoch": 1.99,
"learning_rate": 1.5409470416413943e-06,
"loss": 0.8683,
"step": 71500
},
{
"epoch": 1.99,
"learning_rate": 1.5334815301069374e-06,
"loss": 0.8306,
"step": 71600
},
{
"epoch": 1.99,
"learning_rate": 1.526026138315128e-06,
"loss": 0.8103,
"step": 71700
},
{
"epoch": 2.0,
"learning_rate": 1.5185809443261897e-06,
"loss": 0.8605,
"step": 71800
},
{
"epoch": 2.0,
"learning_rate": 1.5111460260935695e-06,
"loss": 0.833,
"step": 71900
},
{
"epoch": 2.0,
"learning_rate": 1.5037214614631234e-06,
"loss": 0.9071,
"step": 72000
},
{
"epoch": 2.01,
"learning_rate": 1.496307328172306e-06,
"loss": 0.854,
"step": 72100
},
{
"epoch": 2.01,
"learning_rate": 1.4889037038493488e-06,
"loss": 0.8733,
"step": 72200
},
{
"epoch": 2.01,
"learning_rate": 1.4815106660124517e-06,
"loss": 0.8441,
"step": 72300
},
{
"epoch": 2.01,
"learning_rate": 1.4741282920689736e-06,
"loss": 0.874,
"step": 72400
},
{
"epoch": 2.02,
"learning_rate": 1.4667566593146167e-06,
"loss": 0.8514,
"step": 72500
},
{
"epoch": 2.02,
"learning_rate": 1.45939584493262e-06,
"loss": 0.8502,
"step": 72600
},
{
"epoch": 2.02,
"learning_rate": 1.4520459259929527e-06,
"loss": 0.8563,
"step": 72700
},
{
"epoch": 2.02,
"learning_rate": 1.4447069794515e-06,
"loss": 0.8676,
"step": 72800
},
{
"epoch": 2.03,
"learning_rate": 1.437379082149271e-06,
"loss": 0.8202,
"step": 72900
},
{
"epoch": 2.03,
"learning_rate": 1.4300623108115793e-06,
"loss": 0.8378,
"step": 73000
},
{
"epoch": 2.03,
"learning_rate": 1.4227567420472487e-06,
"loss": 0.8805,
"step": 73100
},
{
"epoch": 2.04,
"learning_rate": 1.4154624523478095e-06,
"loss": 0.8568,
"step": 73200
},
{
"epoch": 2.04,
"learning_rate": 1.408179518086694e-06,
"loss": 0.8617,
"step": 73300
},
{
"epoch": 2.04,
"learning_rate": 1.4009080155184407e-06,
"loss": 0.8427,
"step": 73400
},
{
"epoch": 2.04,
"learning_rate": 1.393648020777899e-06,
"loss": 0.8675,
"step": 73500
},
{
"epoch": 2.05,
"learning_rate": 1.3863996098794213e-06,
"loss": 0.8523,
"step": 73600
},
{
"epoch": 2.05,
"learning_rate": 1.3791628587160768e-06,
"loss": 0.8456,
"step": 73700
},
{
"epoch": 2.05,
"learning_rate": 1.3719378430588553e-06,
"loss": 0.8492,
"step": 73800
},
{
"epoch": 2.06,
"learning_rate": 1.3647246385558682e-06,
"loss": 0.8459,
"step": 73900
},
{
"epoch": 2.06,
"learning_rate": 1.3575233207315635e-06,
"loss": 0.8667,
"step": 74000
},
{
"epoch": 2.06,
"learning_rate": 1.3503339649859315e-06,
"loss": 0.8342,
"step": 74100
},
{
"epoch": 2.06,
"learning_rate": 1.3431566465937163e-06,
"loss": 0.8684,
"step": 74200
},
{
"epoch": 2.07,
"learning_rate": 1.3359914407036267e-06,
"loss": 0.8315,
"step": 74300
},
{
"epoch": 2.07,
"learning_rate": 1.3288384223375487e-06,
"loss": 0.8773,
"step": 74400
},
{
"epoch": 2.07,
"learning_rate": 1.3216976663897622e-06,
"loss": 0.8355,
"step": 74500
},
{
"epoch": 2.07,
"learning_rate": 1.3145692476261587e-06,
"loss": 0.8641,
"step": 74600
},
{
"epoch": 2.08,
"learning_rate": 1.3074532406834505e-06,
"loss": 0.8655,
"step": 74700
},
{
"epoch": 2.08,
"learning_rate": 1.300349720068398e-06,
"loss": 0.8502,
"step": 74800
},
{
"epoch": 2.08,
"learning_rate": 1.2932587601570245e-06,
"loss": 0.8529,
"step": 74900
},
{
"epoch": 2.09,
"learning_rate": 1.286180435193839e-06,
"loss": 0.8638,
"step": 75000
},
{
"epoch": 2.09,
"eval_bleu": 8.1623,
"eval_gen_len": 18.979,
"eval_loss": 0.878086507320404,
"eval_runtime": 964.1689,
"eval_samples_per_second": 2.07,
"eval_steps_per_second": 1.035,
"step": 75000
},
{
"epoch": 2.09,
"learning_rate": 1.2791148192910586e-06,
"loss": 0.8571,
"step": 75100
},
{
"epoch": 2.09,
"learning_rate": 1.2720619864278338e-06,
"loss": 0.8588,
"step": 75200
},
{
"epoch": 2.09,
"learning_rate": 1.2650220104494714e-06,
"loss": 0.8413,
"step": 75300
},
{
"epoch": 2.1,
"learning_rate": 1.2579949650666648e-06,
"loss": 0.8627,
"step": 75400
},
{
"epoch": 2.1,
"learning_rate": 1.2509809238547165e-06,
"loss": 0.8522,
"step": 75500
},
{
"epoch": 2.1,
"learning_rate": 1.2439799602527741e-06,
"loss": 0.8369,
"step": 75600
},
{
"epoch": 2.11,
"learning_rate": 1.2369921475630586e-06,
"loss": 0.8676,
"step": 75700
},
{
"epoch": 2.11,
"learning_rate": 1.230017558950096e-06,
"loss": 0.8744,
"step": 75800
},
{
"epoch": 2.11,
"learning_rate": 1.2230562674399538e-06,
"loss": 0.8681,
"step": 75900
},
{
"epoch": 2.11,
"learning_rate": 1.2161083459194714e-06,
"loss": 0.8504,
"step": 76000
},
{
"epoch": 2.12,
"learning_rate": 1.2091738671355039e-06,
"loss": 0.8568,
"step": 76100
},
{
"epoch": 2.12,
"learning_rate": 1.2022529036941546e-06,
"loss": 0.862,
"step": 76200
},
{
"epoch": 2.12,
"learning_rate": 1.1953455280600188e-06,
"loss": 0.8509,
"step": 76300
},
{
"epoch": 2.12,
"learning_rate": 1.188451812555422e-06,
"loss": 0.8718,
"step": 76400
},
{
"epoch": 2.13,
"learning_rate": 1.1815718293596653e-06,
"loss": 0.877,
"step": 76500
},
{
"epoch": 2.13,
"learning_rate": 1.1747056505082658e-06,
"loss": 0.8373,
"step": 76600
},
{
"epoch": 2.13,
"learning_rate": 1.1678533478922075e-06,
"loss": 0.8603,
"step": 76700
},
{
"epoch": 2.14,
"learning_rate": 1.1610149932571847e-06,
"loss": 0.8459,
"step": 76800
},
{
"epoch": 2.14,
"learning_rate": 1.1541906582028526e-06,
"loss": 0.8692,
"step": 76900
},
{
"epoch": 2.14,
"learning_rate": 1.1473804141820783e-06,
"loss": 0.8508,
"step": 77000
},
{
"epoch": 2.14,
"learning_rate": 1.1405843325001878e-06,
"loss": 0.8799,
"step": 77100
},
{
"epoch": 2.15,
"learning_rate": 1.1338024843142265e-06,
"loss": 0.8758,
"step": 77200
},
{
"epoch": 2.15,
"learning_rate": 1.1270349406322109e-06,
"loss": 0.8531,
"step": 77300
},
{
"epoch": 2.15,
"learning_rate": 1.1202817723123807e-06,
"loss": 0.8452,
"step": 77400
},
{
"epoch": 2.16,
"learning_rate": 1.1135430500624675e-06,
"loss": 0.9195,
"step": 77500
},
{
"epoch": 2.16,
"learning_rate": 1.1068188444389444e-06,
"loss": 0.8824,
"step": 77600
},
{
"epoch": 2.16,
"learning_rate": 1.10010922584629e-06,
"loss": 0.8379,
"step": 77700
},
{
"epoch": 2.16,
"learning_rate": 1.0934142645362547e-06,
"loss": 0.8654,
"step": 77800
},
{
"epoch": 2.17,
"learning_rate": 1.0867340306071228e-06,
"loss": 0.8671,
"step": 77900
},
{
"epoch": 2.17,
"learning_rate": 1.0800685940029742e-06,
"loss": 0.8641,
"step": 78000
},
{
"epoch": 2.17,
"learning_rate": 1.073418024512964e-06,
"loss": 0.8477,
"step": 78100
},
{
"epoch": 2.17,
"learning_rate": 1.066782391770576e-06,
"loss": 0.8477,
"step": 78200
},
{
"epoch": 2.18,
"learning_rate": 1.0601617652529066e-06,
"loss": 0.8356,
"step": 78300
},
{
"epoch": 2.18,
"learning_rate": 1.0535562142799325e-06,
"loss": 0.8803,
"step": 78400
},
{
"epoch": 2.18,
"learning_rate": 1.0469658080137801e-06,
"loss": 0.8187,
"step": 78500
},
{
"epoch": 2.19,
"learning_rate": 1.0403906154580127e-06,
"loss": 0.8548,
"step": 78600
},
{
"epoch": 2.19,
"learning_rate": 1.0338307054568977e-06,
"loss": 0.8409,
"step": 78700
},
{
"epoch": 2.19,
"learning_rate": 1.027286146694689e-06,
"loss": 0.8734,
"step": 78800
},
{
"epoch": 2.19,
"learning_rate": 1.0207570076949099e-06,
"loss": 0.858,
"step": 78900
},
{
"epoch": 2.2,
"learning_rate": 1.0142433568196347e-06,
"loss": 0.8373,
"step": 79000
},
{
"epoch": 2.2,
"learning_rate": 1.007745262268768e-06,
"loss": 0.8376,
"step": 79100
},
{
"epoch": 2.2,
"learning_rate": 1.0012627920793424e-06,
"loss": 0.8204,
"step": 79200
},
{
"epoch": 2.21,
"learning_rate": 9.947960141247918e-07,
"loss": 0.8247,
"step": 79300
},
{
"epoch": 2.21,
"learning_rate": 9.883449961142504e-07,
"loss": 0.8753,
"step": 79400
},
{
"epoch": 2.21,
"learning_rate": 9.819098055918424e-07,
"loss": 0.8703,
"step": 79500
},
{
"epoch": 2.21,
"learning_rate": 9.7549050993597e-07,
"loss": 0.823,
"step": 79600
},
{
"epoch": 2.22,
"learning_rate": 9.690871763586137e-07,
"loss": 0.8877,
"step": 79700
},
{
"epoch": 2.22,
"learning_rate": 9.626998719046263e-07,
"loss": 0.8174,
"step": 79800
},
{
"epoch": 2.22,
"learning_rate": 9.563286634510293e-07,
"loss": 0.8523,
"step": 79900
},
{
"epoch": 2.22,
"learning_rate": 9.499736177063165e-07,
"loss": 0.8293,
"step": 80000
},
{
"epoch": 2.22,
"eval_bleu": 8.0989,
"eval_gen_len": 18.98,
"eval_loss": 0.8752478361129761,
"eval_runtime": 964.0098,
"eval_samples_per_second": 2.071,
"eval_steps_per_second": 1.035,
"step": 80000
},
{
"epoch": 2.23,
"learning_rate": 9.436348012097496e-07,
"loss": 0.8633,
"step": 80100
},
{
"epoch": 2.23,
"learning_rate": 9.373122803306672e-07,
"loss": 0.8656,
"step": 80200
},
{
"epoch": 2.23,
"learning_rate": 9.31006121267791e-07,
"loss": 0.8289,
"step": 80300
},
{
"epoch": 2.24,
"learning_rate": 9.247163900485232e-07,
"loss": 0.8325,
"step": 80400
},
{
"epoch": 2.24,
"learning_rate": 9.184431525282659e-07,
"loss": 0.8779,
"step": 80500
},
{
"epoch": 2.24,
"learning_rate": 9.121864743897266e-07,
"loss": 0.8456,
"step": 80600
},
{
"epoch": 2.24,
"learning_rate": 9.059464211422286e-07,
"loss": 0.8431,
"step": 80700
},
{
"epoch": 2.25,
"learning_rate": 8.9972305812103e-07,
"loss": 0.8287,
"step": 80800
},
{
"epoch": 2.25,
"learning_rate": 8.935164504866367e-07,
"loss": 0.842,
"step": 80900
},
{
"epoch": 2.25,
"learning_rate": 8.873266632241201e-07,
"loss": 0.8748,
"step": 81000
},
{
"epoch": 2.26,
"learning_rate": 8.811537611424383e-07,
"loss": 0.8652,
"step": 81100
},
{
"epoch": 2.26,
"learning_rate": 8.749978088737541e-07,
"loss": 0.8327,
"step": 81200
},
{
"epoch": 2.26,
"learning_rate": 8.688588708727621e-07,
"loss": 0.8433,
"step": 81300
},
{
"epoch": 2.26,
"learning_rate": 8.627370114160133e-07,
"loss": 0.861,
"step": 81400
},
{
"epoch": 2.27,
"learning_rate": 8.566322946012389e-07,
"loss": 0.852,
"step": 81500
},
{
"epoch": 2.27,
"learning_rate": 8.505447843466836e-07,
"loss": 0.8334,
"step": 81600
},
{
"epoch": 2.27,
"learning_rate": 8.444745443904337e-07,
"loss": 0.8509,
"step": 81700
},
{
"epoch": 2.27,
"learning_rate": 8.384216382897476e-07,
"loss": 0.8272,
"step": 81800
},
{
"epoch": 2.28,
"learning_rate": 8.323861294203964e-07,
"loss": 0.7921,
"step": 81900
},
{
"epoch": 2.28,
"learning_rate": 8.263680809759955e-07,
"loss": 0.8153,
"step": 82000
},
{
"epoch": 2.28,
"learning_rate": 8.203675559673441e-07,
"loss": 0.8168,
"step": 82100
},
{
"epoch": 2.29,
"learning_rate": 8.143846172217671e-07,
"loss": 0.872,
"step": 82200
},
{
"epoch": 2.29,
"learning_rate": 8.084193273824531e-07,
"loss": 0.8519,
"step": 82300
},
{
"epoch": 2.29,
"learning_rate": 8.024717489078032e-07,
"loss": 0.8594,
"step": 82400
},
{
"epoch": 2.29,
"learning_rate": 7.965419440707756e-07,
"loss": 0.8445,
"step": 82500
},
{
"epoch": 2.3,
"learning_rate": 7.90629974958232e-07,
"loss": 0.8429,
"step": 82600
},
{
"epoch": 2.3,
"learning_rate": 7.84735903470289e-07,
"loss": 0.819,
"step": 82700
},
{
"epoch": 2.3,
"learning_rate": 7.788597913196702e-07,
"loss": 0.829,
"step": 82800
},
{
"epoch": 2.31,
"learning_rate": 7.730017000310575e-07,
"loss": 0.8163,
"step": 82900
},
{
"epoch": 2.31,
"learning_rate": 7.671616909404508e-07,
"loss": 0.8699,
"step": 83000
},
{
"epoch": 2.31,
"learning_rate": 7.613398251945239e-07,
"loss": 0.8303,
"step": 83100
},
{
"epoch": 2.31,
"learning_rate": 7.555361637499833e-07,
"loss": 0.8012,
"step": 83200
},
{
"epoch": 2.32,
"learning_rate": 7.49750767372932e-07,
"loss": 0.8178,
"step": 83300
},
{
"epoch": 2.32,
"learning_rate": 7.439836966382303e-07,
"loss": 0.847,
"step": 83400
},
{
"epoch": 2.32,
"learning_rate": 7.382350119288647e-07,
"loss": 0.8623,
"step": 83500
},
{
"epoch": 2.32,
"learning_rate": 7.325047734353155e-07,
"loss": 0.8079,
"step": 83600
},
{
"epoch": 2.33,
"learning_rate": 7.267930411549234e-07,
"loss": 0.8337,
"step": 83700
},
{
"epoch": 2.33,
"learning_rate": 7.210998748912657e-07,
"loss": 0.8399,
"step": 83800
},
{
"epoch": 2.33,
"learning_rate": 7.154253342535248e-07,
"loss": 0.851,
"step": 83900
},
{
"epoch": 2.34,
"learning_rate": 7.097694786558693e-07,
"loss": 0.8304,
"step": 84000
},
{
"epoch": 2.34,
"learning_rate": 7.041323673168307e-07,
"loss": 0.846,
"step": 84100
},
{
"epoch": 2.34,
"learning_rate": 6.985140592586781e-07,
"loss": 0.8835,
"step": 84200
},
{
"epoch": 2.34,
"learning_rate": 6.929146133068102e-07,
"loss": 0.8588,
"step": 84300
},
{
"epoch": 2.35,
"learning_rate": 6.873340880891308e-07,
"loss": 0.8275,
"step": 84400
},
{
"epoch": 2.35,
"learning_rate": 6.817725420354365e-07,
"loss": 0.8529,
"step": 84500
},
{
"epoch": 2.35,
"learning_rate": 6.762300333768082e-07,
"loss": 0.8705,
"step": 84600
},
{
"epoch": 2.36,
"learning_rate": 6.707066201450003e-07,
"loss": 0.8591,
"step": 84700
},
{
"epoch": 2.36,
"learning_rate": 6.652023601718282e-07,
"loss": 0.8235,
"step": 84800
},
{
"epoch": 2.36,
"learning_rate": 6.597173110885732e-07,
"loss": 0.8575,
"step": 84900
},
{
"epoch": 2.36,
"learning_rate": 6.542515303253666e-07,
"loss": 0.8625,
"step": 85000
},
{
"epoch": 2.36,
"eval_bleu": 8.176,
"eval_gen_len": 18.979,
"eval_loss": 0.8743442296981812,
"eval_runtime": 965.6565,
"eval_samples_per_second": 2.067,
"eval_steps_per_second": 1.033,
"step": 85000
},
{
"epoch": 2.37,
"learning_rate": 6.488050751105979e-07,
"loss": 0.8735,
"step": 85100
},
{
"epoch": 2.37,
"learning_rate": 6.433780024703124e-07,
"loss": 0.8721,
"step": 85200
},
{
"epoch": 2.37,
"learning_rate": 6.379703692276104e-07,
"loss": 0.8554,
"step": 85300
},
{
"epoch": 2.37,
"learning_rate": 6.325822320020608e-07,
"loss": 0.8569,
"step": 85400
},
{
"epoch": 2.38,
"learning_rate": 6.272136472091008e-07,
"loss": 0.8107,
"step": 85500
},
{
"epoch": 2.38,
"learning_rate": 6.218646710594465e-07,
"loss": 0.8671,
"step": 85600
},
{
"epoch": 2.38,
"learning_rate": 6.165353595585069e-07,
"loss": 0.8392,
"step": 85700
},
{
"epoch": 2.39,
"learning_rate": 6.112257685057973e-07,
"loss": 0.8353,
"step": 85800
},
{
"epoch": 2.39,
"learning_rate": 6.059359534943501e-07,
"loss": 0.8244,
"step": 85900
},
{
"epoch": 2.39,
"learning_rate": 6.006659699101419e-07,
"loss": 0.8328,
"step": 86000
},
{
"epoch": 2.39,
"learning_rate": 5.954158729315032e-07,
"loss": 0.8333,
"step": 86100
},
{
"epoch": 2.4,
"learning_rate": 5.901857175285488e-07,
"loss": 0.8651,
"step": 86200
},
{
"epoch": 2.4,
"learning_rate": 5.849755584625985e-07,
"loss": 0.8671,
"step": 86300
},
{
"epoch": 2.4,
"learning_rate": 5.797854502856029e-07,
"loss": 0.8453,
"step": 86400
},
{
"epoch": 2.41,
"learning_rate": 5.746154473395752e-07,
"loss": 0.8358,
"step": 86500
},
{
"epoch": 2.41,
"learning_rate": 5.694656037560206e-07,
"loss": 0.8291,
"step": 86600
},
{
"epoch": 2.41,
"learning_rate": 5.643359734553693e-07,
"loss": 0.8541,
"step": 86700
},
{
"epoch": 2.41,
"learning_rate": 5.592266101464122e-07,
"loss": 0.8466,
"step": 86800
},
{
"epoch": 2.42,
"learning_rate": 5.541375673257394e-07,
"loss": 0.8221,
"step": 86900
},
{
"epoch": 2.42,
"learning_rate": 5.490688982771769e-07,
"loss": 0.8347,
"step": 87000
},
{
"epoch": 2.42,
"learning_rate": 5.440206560712352e-07,
"loss": 0.8715,
"step": 87100
},
{
"epoch": 2.42,
"learning_rate": 5.389928935645452e-07,
"loss": 0.854,
"step": 87200
},
{
"epoch": 2.43,
"learning_rate": 5.339856633993124e-07,
"loss": 0.8747,
"step": 87300
},
{
"epoch": 2.43,
"learning_rate": 5.28999018002761e-07,
"loss": 0.8586,
"step": 87400
},
{
"epoch": 2.43,
"learning_rate": 5.240330095865856e-07,
"loss": 0.8247,
"step": 87500
},
{
"epoch": 2.44,
"learning_rate": 5.190876901464067e-07,
"loss": 0.8357,
"step": 87600
},
{
"epoch": 2.44,
"learning_rate": 5.14163111461225e-07,
"loss": 0.847,
"step": 87700
},
{
"epoch": 2.44,
"learning_rate": 5.092593250928782e-07,
"loss": 0.8241,
"step": 87800
},
{
"epoch": 2.44,
"learning_rate": 5.043763823855036e-07,
"loss": 0.8706,
"step": 87900
},
{
"epoch": 2.45,
"learning_rate": 4.995143344649964e-07,
"loss": 0.8625,
"step": 88000
},
{
"epoch": 2.45,
"learning_rate": 4.946732322384795e-07,
"loss": 0.8476,
"step": 88100
},
{
"epoch": 2.45,
"learning_rate": 4.89853126393767e-07,
"loss": 0.8344,
"step": 88200
},
{
"epoch": 2.46,
"learning_rate": 4.850540673988346e-07,
"loss": 0.8456,
"step": 88300
},
{
"epoch": 2.46,
"learning_rate": 4.802761055012914e-07,
"loss": 0.8175,
"step": 88400
},
{
"epoch": 2.46,
"learning_rate": 4.755192907278536e-07,
"loss": 0.848,
"step": 88500
},
{
"epoch": 2.46,
"learning_rate": 4.7078367288381886e-07,
"loss": 0.8592,
"step": 88600
},
{
"epoch": 2.47,
"learning_rate": 4.660693015525486e-07,
"loss": 0.82,
"step": 88700
},
{
"epoch": 2.47,
"learning_rate": 4.613762260949456e-07,
"loss": 0.8745,
"step": 88800
},
{
"epoch": 2.47,
"learning_rate": 4.567044956489394e-07,
"loss": 0.8827,
"step": 88900
},
{
"epoch": 2.48,
"learning_rate": 4.520541591289701e-07,
"loss": 0.862,
"step": 89000
},
{
"epoch": 2.48,
"learning_rate": 4.4742526522547626e-07,
"loss": 0.8684,
"step": 89100
},
{
"epoch": 2.48,
"learning_rate": 4.428178624043866e-07,
"loss": 0.8537,
"step": 89200
},
{
"epoch": 2.48,
"learning_rate": 4.382319989066117e-07,
"loss": 0.8451,
"step": 89300
},
{
"epoch": 2.49,
"learning_rate": 4.336677227475383e-07,
"loss": 0.8589,
"step": 89400
},
{
"epoch": 2.49,
"learning_rate": 4.2912508171652765e-07,
"loss": 0.8102,
"step": 89500
},
{
"epoch": 2.49,
"learning_rate": 4.2460412337641504e-07,
"loss": 0.8662,
"step": 89600
},
{
"epoch": 2.49,
"learning_rate": 4.2010489506300933e-07,
"loss": 0.8566,
"step": 89700
},
{
"epoch": 2.5,
"learning_rate": 4.156274438846017e-07,
"loss": 0.8663,
"step": 89800
},
{
"epoch": 2.5,
"learning_rate": 4.11171816721469e-07,
"loss": 0.8125,
"step": 89900
},
{
"epoch": 2.5,
"learning_rate": 4.0673806022538425e-07,
"loss": 0.8605,
"step": 90000
},
{
"epoch": 2.5,
"eval_bleu": 8.0117,
"eval_gen_len": 18.9805,
"eval_loss": 0.8721033334732056,
"eval_runtime": 963.8746,
"eval_samples_per_second": 2.071,
"eval_steps_per_second": 1.035,
"step": 90000
},
{
"epoch": 2.51,
"learning_rate": 4.023262208191284e-07,
"loss": 0.8296,
"step": 90100
},
{
"epoch": 2.51,
"learning_rate": 3.9793634469600216e-07,
"loss": 0.83,
"step": 90200
},
{
"epoch": 2.51,
"learning_rate": 3.9356847781934575e-07,
"loss": 0.8432,
"step": 90300
},
{
"epoch": 2.51,
"learning_rate": 3.892226659220552e-07,
"loss": 0.8396,
"step": 90400
},
{
"epoch": 2.52,
"learning_rate": 3.8489895450610407e-07,
"loss": 0.8742,
"step": 90500
},
{
"epoch": 2.52,
"learning_rate": 3.8059738884206775e-07,
"loss": 0.8124,
"step": 90600
},
{
"epoch": 2.52,
"learning_rate": 3.7631801396864757e-07,
"loss": 0.8238,
"step": 90700
},
{
"epoch": 2.53,
"learning_rate": 3.7206087469220195e-07,
"loss": 0.8443,
"step": 90800
},
{
"epoch": 2.53,
"learning_rate": 3.6782601558627563e-07,
"loss": 0.8567,
"step": 90900
},
{
"epoch": 2.53,
"learning_rate": 3.6361348099113123e-07,
"loss": 0.8406,
"step": 91000
},
{
"epoch": 2.53,
"learning_rate": 3.5942331501329003e-07,
"loss": 0.8667,
"step": 91100
},
{
"epoch": 2.54,
"learning_rate": 3.552555615250658e-07,
"loss": 0.8206,
"step": 91200
},
{
"epoch": 2.54,
"learning_rate": 3.511102641641051e-07,
"loss": 0.8545,
"step": 91300
},
{
"epoch": 2.54,
"learning_rate": 3.469874663329342e-07,
"loss": 0.8578,
"step": 91400
},
{
"epoch": 2.54,
"learning_rate": 3.428872111985021e-07,
"loss": 0.8555,
"step": 91500
},
{
"epoch": 2.55,
"learning_rate": 3.388095416917267e-07,
"loss": 0.8161,
"step": 91600
},
{
"epoch": 2.55,
"learning_rate": 3.3475450050705125e-07,
"loss": 0.8234,
"step": 91700
},
{
"epoch": 2.55,
"learning_rate": 3.3072213010199053e-07,
"loss": 0.8342,
"step": 91800
},
{
"epoch": 2.56,
"learning_rate": 3.267124726966903e-07,
"loss": 0.8458,
"step": 91900
},
{
"epoch": 2.56,
"learning_rate": 3.2272557027348524e-07,
"loss": 0.8573,
"step": 92000
},
{
"epoch": 2.56,
"learning_rate": 3.187614645764564e-07,
"loss": 0.8544,
"step": 92100
},
{
"epoch": 2.56,
"learning_rate": 3.1482019711099735e-07,
"loss": 0.8455,
"step": 92200
},
{
"epoch": 2.57,
"learning_rate": 3.109018091433802e-07,
"loss": 0.847,
"step": 92300
},
{
"epoch": 2.57,
"learning_rate": 3.07006341700318e-07,
"loss": 0.8277,
"step": 92400
},
{
"epoch": 2.57,
"learning_rate": 3.031338355685418e-07,
"loss": 0.8297,
"step": 92500
},
{
"epoch": 2.58,
"learning_rate": 2.992843312943702e-07,
"loss": 0.8648,
"step": 92600
},
{
"epoch": 2.58,
"learning_rate": 2.954578691832835e-07,
"loss": 0.8469,
"step": 92700
},
{
"epoch": 2.58,
"learning_rate": 2.9165448929950685e-07,
"loss": 0.8757,
"step": 92800
},
{
"epoch": 2.58,
"learning_rate": 2.878742314655844e-07,
"loss": 0.8526,
"step": 92900
},
{
"epoch": 2.59,
"learning_rate": 2.8411713526196677e-07,
"loss": 0.8369,
"step": 93000
},
{
"epoch": 2.59,
"learning_rate": 2.80383240026596e-07,
"loss": 0.8428,
"step": 93100
},
{
"epoch": 2.59,
"learning_rate": 2.766725848544907e-07,
"loss": 0.8428,
"step": 93200
},
{
"epoch": 2.59,
"learning_rate": 2.7298520859734054e-07,
"loss": 0.8695,
"step": 93300
},
{
"epoch": 2.6,
"learning_rate": 2.6932114986309874e-07,
"loss": 0.8534,
"step": 93400
},
{
"epoch": 2.6,
"learning_rate": 2.6568044701557494e-07,
"loss": 0.8515,
"step": 93500
},
{
"epoch": 2.6,
"learning_rate": 2.6206313817403627e-07,
"loss": 0.8642,
"step": 93600
},
{
"epoch": 2.61,
"learning_rate": 2.5846926121280843e-07,
"loss": 0.8184,
"step": 93700
},
{
"epoch": 2.61,
"learning_rate": 2.5489885376087626e-07,
"loss": 0.8173,
"step": 93800
},
{
"epoch": 2.61,
"learning_rate": 2.5135195320149355e-07,
"loss": 0.85,
"step": 93900
},
{
"epoch": 2.61,
"learning_rate": 2.478285966717889e-07,
"loss": 0.858,
"step": 94000
},
{
"epoch": 2.62,
"learning_rate": 2.4432882106237786e-07,
"loss": 0.8104,
"step": 94100
},
{
"epoch": 2.62,
"learning_rate": 2.408526630169772e-07,
"loss": 0.8469,
"step": 94200
},
{
"epoch": 2.62,
"learning_rate": 2.3740015893201906e-07,
"loss": 0.8599,
"step": 94300
},
{
"epoch": 2.63,
"learning_rate": 2.3397134495627278e-07,
"loss": 0.874,
"step": 94400
},
{
"epoch": 2.63,
"learning_rate": 2.305662569904646e-07,
"loss": 0.8251,
"step": 94500
},
{
"epoch": 2.63,
"learning_rate": 2.2718493068690229e-07,
"loss": 0.86,
"step": 94600
},
{
"epoch": 2.63,
"learning_rate": 2.2382740144910236e-07,
"loss": 0.8766,
"step": 94700
},
{
"epoch": 2.64,
"learning_rate": 2.2049370443141704e-07,
"loss": 0.8541,
"step": 94800
},
{
"epoch": 2.64,
"learning_rate": 2.171838745386695e-07,
"loss": 0.8276,
"step": 94900
},
{
"epoch": 2.64,
"learning_rate": 2.1389794642578649e-07,
"loss": 0.8479,
"step": 95000
},
{
"epoch": 2.64,
"eval_bleu": 8.1008,
"eval_gen_len": 18.978,
"eval_loss": 0.8710653781890869,
"eval_runtime": 966.0685,
"eval_samples_per_second": 2.066,
"eval_steps_per_second": 1.033,
"step": 95000
},
{
"epoch": 2.64,
"learning_rate": 2.1063595449743595e-07,
"loss": 0.8244,
"step": 95100
},
{
"epoch": 2.65,
"learning_rate": 2.073979329076664e-07,
"loss": 0.8897,
"step": 95200
},
{
"epoch": 2.65,
"learning_rate": 2.0418391555955042e-07,
"loss": 0.8269,
"step": 95300
},
{
"epoch": 2.65,
"learning_rate": 2.0099393610482688e-07,
"loss": 0.8353,
"step": 95400
},
{
"epoch": 2.66,
"learning_rate": 1.9782802794355239e-07,
"loss": 0.8267,
"step": 95500
},
{
"epoch": 2.66,
"learning_rate": 1.9468622422374962e-07,
"loss": 0.8484,
"step": 95600
},
{
"epoch": 2.66,
"learning_rate": 1.9156855784106004e-07,
"loss": 0.8665,
"step": 95700
},
{
"epoch": 2.66,
"learning_rate": 1.8847506143839983e-07,
"loss": 0.8377,
"step": 95800
},
{
"epoch": 2.67,
"learning_rate": 1.854057674056181e-07,
"loss": 0.859,
"step": 95900
},
{
"epoch": 2.67,
"learning_rate": 1.8236070787915754e-07,
"loss": 0.8135,
"step": 96000
},
{
"epoch": 2.67,
"learning_rate": 1.793399147417188e-07,
"loss": 0.8111,
"step": 96100
},
{
"epoch": 2.68,
"learning_rate": 1.7634341962192602e-07,
"loss": 0.8239,
"step": 96200
},
{
"epoch": 2.68,
"learning_rate": 1.7337125389399496e-07,
"loss": 0.8728,
"step": 96300
},
{
"epoch": 2.68,
"learning_rate": 1.7042344867740645e-07,
"loss": 0.8582,
"step": 96400
},
{
"epoch": 2.68,
"learning_rate": 1.6750003483657739e-07,
"loss": 0.8271,
"step": 96500
},
{
"epoch": 2.69,
"learning_rate": 1.646010429805417e-07,
"loss": 0.8376,
"step": 96600
},
{
"epoch": 2.69,
"learning_rate": 1.6172650346262642e-07,
"loss": 0.867,
"step": 96700
},
{
"epoch": 2.69,
"learning_rate": 1.5887644638013566e-07,
"loss": 0.8924,
"step": 96800
},
{
"epoch": 2.69,
"learning_rate": 1.5605090157403495e-07,
"loss": 0.8466,
"step": 96900
},
{
"epoch": 2.7,
"learning_rate": 1.5324989862863792e-07,
"loss": 0.8462,
"step": 97000
},
{
"epoch": 2.7,
"learning_rate": 1.5047346687129898e-07,
"loss": 0.8679,
"step": 97100
},
{
"epoch": 2.7,
"learning_rate": 1.4772163537210393e-07,
"loss": 0.8395,
"step": 97200
},
{
"epoch": 2.71,
"learning_rate": 1.4499443294356541e-07,
"loss": 0.8514,
"step": 97300
},
{
"epoch": 2.71,
"learning_rate": 1.422918881403243e-07,
"loss": 0.8532,
"step": 97400
},
{
"epoch": 2.71,
"learning_rate": 1.3961402925884744e-07,
"loss": 0.8172,
"step": 97500
},
{
"epoch": 2.71,
"learning_rate": 1.36960884337132e-07,
"loss": 0.7917,
"step": 97600
},
{
"epoch": 2.72,
"learning_rate": 1.3433248115441362e-07,
"loss": 0.8305,
"step": 97700
},
{
"epoch": 2.72,
"learning_rate": 1.3172884723087336e-07,
"loss": 0.8496,
"step": 97800
},
{
"epoch": 2.72,
"learning_rate": 1.2915000982735155e-07,
"loss": 0.8685,
"step": 97900
},
{
"epoch": 2.73,
"learning_rate": 1.2659599594506106e-07,
"loss": 0.8706,
"step": 98000
},
{
"epoch": 2.73,
"learning_rate": 1.2406683232530414e-07,
"loss": 0.8015,
"step": 98100
},
{
"epoch": 2.73,
"learning_rate": 1.2156254544919476e-07,
"loss": 0.8639,
"step": 98200
},
{
"epoch": 2.73,
"learning_rate": 1.1908316153737858e-07,
"loss": 0.8624,
"step": 98300
},
{
"epoch": 2.74,
"learning_rate": 1.1662870654975955e-07,
"loss": 0.8271,
"step": 98400
},
{
"epoch": 2.74,
"learning_rate": 1.1419920618522984e-07,
"loss": 0.8196,
"step": 98500
},
{
"epoch": 2.74,
"learning_rate": 1.1179468588139702e-07,
"loss": 0.8528,
"step": 98600
},
{
"epoch": 2.74,
"learning_rate": 1.0941517081432096e-07,
"loss": 0.8642,
"step": 98700
},
{
"epoch": 2.75,
"learning_rate": 1.0706068589824925e-07,
"loss": 0.8515,
"step": 98800
},
{
"epoch": 2.75,
"learning_rate": 1.0473125578535526e-07,
"loss": 0.8398,
"step": 98900
},
{
"epoch": 2.75,
"learning_rate": 1.0242690486548134e-07,
"loss": 0.8797,
"step": 99000
},
{
"epoch": 2.76,
"learning_rate": 1.0014765726588437e-07,
"loss": 0.8537,
"step": 99100
},
{
"epoch": 2.76,
"learning_rate": 9.789353685097953e-08,
"loss": 0.812,
"step": 99200
},
{
"epoch": 2.76,
"learning_rate": 9.566456722209432e-08,
"loss": 0.8588,
"step": 99300
},
{
"epoch": 2.76,
"learning_rate": 9.34607717172195e-08,
"loss": 0.8394,
"step": 99400
},
{
"epoch": 2.77,
"learning_rate": 9.12821734107644e-08,
"loss": 0.8493,
"step": 99500
},
{
"epoch": 2.77,
"learning_rate": 8.91287951133174e-08,
"loss": 0.8376,
"step": 99600
},
{
"epoch": 2.77,
"learning_rate": 8.700065937140401e-08,
"loss": 0.832,
"step": 99700
},
{
"epoch": 2.78,
"learning_rate": 8.489778846725417e-08,
"loss": 0.8802,
"step": 99800
},
{
"epoch": 2.78,
"learning_rate": 8.282020441856637e-08,
"loss": 0.8084,
"step": 99900
},
{
"epoch": 2.78,
"learning_rate": 8.076792897827757e-08,
"loss": 0.8391,
"step": 100000
},
{
"epoch": 2.78,
"eval_bleu": 8.2041,
"eval_gen_len": 18.9795,
"eval_loss": 0.8708174228668213,
"eval_runtime": 963.8453,
"eval_samples_per_second": 2.071,
"eval_steps_per_second": 1.035,
"step": 100000
},
{
"epoch": 2.78,
"learning_rate": 7.874098363433668e-08,
"loss": 0.8315,
"step": 100100
},
{
"epoch": 2.79,
"learning_rate": 7.673938960947924e-08,
"loss": 0.8792,
"step": 100200
},
{
"epoch": 2.79,
"learning_rate": 7.476316786100336e-08,
"loss": 0.8167,
"step": 100300
},
{
"epoch": 2.79,
"learning_rate": 7.281233908055269e-08,
"loss": 0.8536,
"step": 100400
},
{
"epoch": 2.79,
"learning_rate": 7.088692369389888e-08,
"loss": 0.8216,
"step": 100500
},
{
"epoch": 2.8,
"learning_rate": 6.89869418607264e-08,
"loss": 0.8033,
"step": 100600
},
{
"epoch": 2.8,
"learning_rate": 6.711241347442415e-08,
"loss": 0.8184,
"step": 100700
},
{
"epoch": 2.8,
"learning_rate": 6.526335816187474e-08,
"loss": 0.847,
"step": 100800
},
{
"epoch": 2.81,
"learning_rate": 6.34397952832505e-08,
"loss": 0.8446,
"step": 100900
},
{
"epoch": 2.81,
"learning_rate": 6.164174393181038e-08,
"loss": 0.869,
"step": 101000
},
{
"epoch": 2.81,
"learning_rate": 5.986922293369834e-08,
"loss": 0.8355,
"step": 101100
},
{
"epoch": 2.81,
"learning_rate": 5.812225084774969e-08,
"loss": 0.8527,
"step": 101200
},
{
"epoch": 2.82,
"learning_rate": 5.640084596529399e-08,
"loss": 0.8716,
"step": 101300
},
{
"epoch": 2.82,
"learning_rate": 5.4705026309964104e-08,
"loss": 0.8185,
"step": 101400
},
{
"epoch": 2.82,
"learning_rate": 5.3034809637508846e-08,
"loss": 0.8322,
"step": 101500
},
{
"epoch": 2.83,
"learning_rate": 5.139021343560452e-08,
"loss": 0.81,
"step": 101600
},
{
"epoch": 2.83,
"learning_rate": 4.977125492367452e-08,
"loss": 0.8391,
"step": 101700
},
{
"epoch": 2.83,
"learning_rate": 4.817795105270723e-08,
"loss": 0.8142,
"step": 101800
},
{
"epoch": 2.83,
"learning_rate": 4.661031850507924e-08,
"loss": 0.8039,
"step": 101900
},
{
"epoch": 2.84,
"learning_rate": 4.5068373694380775e-08,
"loss": 0.828,
"step": 102000
},
{
"epoch": 2.84,
"learning_rate": 4.355213276524356e-08,
"loss": 0.8512,
"step": 102100
},
{
"epoch": 2.84,
"learning_rate": 4.206161159317129e-08,
"loss": 0.839,
"step": 102200
},
{
"epoch": 2.84,
"learning_rate": 4.059682578437474e-08,
"loss": 0.8535,
"step": 102300
},
{
"epoch": 2.85,
"learning_rate": 3.915779067560743e-08,
"loss": 0.8548,
"step": 102400
},
{
"epoch": 2.85,
"learning_rate": 3.774452133400469e-08,
"loss": 0.8437,
"step": 102500
},
{
"epoch": 2.85,
"learning_rate": 3.635703255692735e-08,
"loss": 0.8258,
"step": 102600
},
{
"epoch": 2.86,
"learning_rate": 3.4995338871804954e-08,
"loss": 0.8327,
"step": 102700
},
{
"epoch": 2.86,
"learning_rate": 3.3659454535985015e-08,
"loss": 0.8453,
"step": 102800
},
{
"epoch": 2.86,
"learning_rate": 3.234939353658345e-08,
"loss": 0.8719,
"step": 102900
},
{
"epoch": 2.86,
"learning_rate": 3.1065169590337453e-08,
"loss": 0.8516,
"step": 103000
},
{
"epoch": 2.87,
"learning_rate": 2.98067961434631e-08,
"loss": 0.8355,
"step": 103100
},
{
"epoch": 2.87,
"learning_rate": 2.857428637151327e-08,
"loss": 0.8453,
"step": 103200
},
{
"epoch": 2.87,
"learning_rate": 2.7367653179240783e-08,
"loss": 0.8211,
"step": 103300
},
{
"epoch": 2.88,
"learning_rate": 2.6186909200462128e-08,
"loss": 0.8567,
"step": 103400
},
{
"epoch": 2.88,
"learning_rate": 2.503206679792647e-08,
"loss": 0.8693,
"step": 103500
},
{
"epoch": 2.88,
"learning_rate": 2.390313806318545e-08,
"loss": 0.8263,
"step": 103600
},
{
"epoch": 2.88,
"learning_rate": 2.2800134816466647e-08,
"loss": 0.8622,
"step": 103700
},
{
"epoch": 2.89,
"learning_rate": 2.172306860654977e-08,
"loss": 0.8541,
"step": 103800
},
{
"epoch": 2.89,
"learning_rate": 2.0671950710645928e-08,
"loss": 0.8408,
"step": 103900
},
{
"epoch": 2.89,
"learning_rate": 1.9646792134279667e-08,
"loss": 0.8284,
"step": 104000
},
{
"epoch": 2.89,
"learning_rate": 1.864760361117296e-08,
"loss": 0.8354,
"step": 104100
},
{
"epoch": 2.9,
"learning_rate": 1.7674395603134442e-08,
"loss": 0.8895,
"step": 104200
},
{
"epoch": 2.9,
"learning_rate": 1.6727178299948133e-08,
"loss": 0.8532,
"step": 104300
},
{
"epoch": 2.9,
"learning_rate": 1.5805961619267396e-08,
"loss": 0.8279,
"step": 104400
},
{
"epoch": 2.91,
"learning_rate": 1.491075520651142e-08,
"loss": 0.8415,
"step": 104500
},
{
"epoch": 2.91,
"learning_rate": 1.4041568434764175e-08,
"loss": 0.8381,
"step": 104600
},
{
"epoch": 2.91,
"learning_rate": 1.3198410404675066e-08,
"loss": 0.8357,
"step": 104700
},
{
"epoch": 2.91,
"learning_rate": 1.2381289944366492e-08,
"loss": 0.8405,
"step": 104800
},
{
"epoch": 2.92,
"learning_rate": 1.1590215609337264e-08,
"loss": 0.8565,
"step": 104900
},
{
"epoch": 2.92,
"learning_rate": 1.0825195682377387e-08,
"loss": 0.8649,
"step": 105000
},
{
"epoch": 2.92,
"eval_bleu": 8.1488,
"eval_gen_len": 18.9785,
"eval_loss": 0.8710347414016724,
"eval_runtime": 973.1045,
"eval_samples_per_second": 2.051,
"eval_steps_per_second": 1.026,
"step": 105000
},
{
"epoch": 2.92,
"learning_rate": 1.0086238173478146e-08,
"loss": 0.8551,
"step": 105100
},
{
"epoch": 2.93,
"learning_rate": 9.373350819749382e-09,
"loss": 0.8313,
"step": 105200
},
{
"epoch": 2.93,
"learning_rate": 8.686541085339006e-09,
"loss": 0.8401,
"step": 105300
},
{
"epoch": 2.93,
"learning_rate": 8.025816161353895e-09,
"loss": 0.8806,
"step": 105400
},
{
"epoch": 2.93,
"learning_rate": 7.391182965785504e-09,
"loss": 0.8271,
"step": 105500
},
{
"epoch": 2.94,
"learning_rate": 6.782648143436321e-09,
"loss": 0.8348,
"step": 105600
},
{
"epoch": 2.94,
"learning_rate": 6.200218065851304e-09,
"loss": 0.8564,
"step": 105700
},
{
"epoch": 2.94,
"learning_rate": 5.6438988312504385e-09,
"loss": 0.8111,
"step": 105800
},
{
"epoch": 2.95,
"learning_rate": 5.113696264466006e-09,
"loss": 0.856,
"step": 105900
},
{
"epoch": 2.95,
"learning_rate": 4.6096159168798616e-09,
"loss": 0.814,
"step": 106000
},
{
"epoch": 2.95,
"learning_rate": 4.1316630663670864e-09,
"loss": 0.887,
"step": 106100
},
{
"epoch": 2.95,
"learning_rate": 3.6798427172390904e-09,
"loss": 0.8207,
"step": 106200
},
{
"epoch": 2.96,
"learning_rate": 3.254159600192819e-09,
"loss": 0.8367,
"step": 106300
},
{
"epoch": 2.96,
"learning_rate": 2.8546181722599597e-09,
"loss": 0.881,
"step": 106400
},
{
"epoch": 2.96,
"learning_rate": 2.48122261676087e-09,
"loss": 0.8333,
"step": 106500
},
{
"epoch": 2.96,
"learning_rate": 2.1339768432609988e-09,
"loss": 0.8364,
"step": 106600
},
{
"epoch": 2.97,
"learning_rate": 1.8128844875289764e-09,
"loss": 0.8306,
"step": 106700
},
{
"epoch": 2.97,
"learning_rate": 1.517948911499978e-09,
"loss": 0.8722,
"step": 106800
},
{
"epoch": 2.97,
"learning_rate": 1.2491732032385295e-09,
"loss": 0.8287,
"step": 106900
},
{
"epoch": 2.98,
"learning_rate": 1.0065601769088106e-09,
"loss": 0.8401,
"step": 107000
},
{
"epoch": 2.98,
"learning_rate": 7.901123727427351e-10,
"loss": 0.8713,
"step": 107100
},
{
"epoch": 2.98,
"learning_rate": 5.998320570149708e-10,
"loss": 0.8514,
"step": 107200
},
{
"epoch": 2.98,
"learning_rate": 4.357212220182372e-10,
"loss": 0.8462,
"step": 107300
},
{
"epoch": 2.99,
"learning_rate": 2.9778158604276684e-10,
"loss": 0.8381,
"step": 107400
},
{
"epoch": 2.99,
"learning_rate": 1.860145933585411e-10,
"loss": 0.8647,
"step": 107500
},
{
"epoch": 2.99,
"learning_rate": 1.0042141420030238e-10,
"loss": 0.8551,
"step": 107600
},
{
"epoch": 3.0,
"learning_rate": 4.1002944754509055e-11,
"loss": 0.8085,
"step": 107700
},
{
"epoch": 3.0,
"learning_rate": 7.759807150731214e-12,
"loss": 0.8357,
"step": 107800
},
{
"epoch": 3.0,
"step": 107877,
"total_flos": 3.211633985499169e+18,
"train_loss": 0.3135226284782316,
"train_runtime": 125809.459,
"train_samples_per_second": 3.43,
"train_steps_per_second": 0.857
}
],
"logging_steps": 100,
"max_steps": 107877,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 2000,
"total_flos": 3.211633985499169e+18,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}