cecilemacaire's picture
Upload 12 files
7629fdf verified
raw
history blame contribute delete
No virus
174 kB
{
"best_metric": 0.15202689170837402,
"best_model_checkpoint": "checkpoints_commonvoice/checkpoint-131848",
"epoch": 40.0,
"global_step": 659240,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03,
"learning_rate": 1.9984831017535344e-05,
"loss": 1.727,
"step": 500
},
{
"epoch": 0.06,
"learning_rate": 1.996969237303562e-05,
"loss": 0.9353,
"step": 1000
},
{
"epoch": 0.09,
"learning_rate": 1.995452339057096e-05,
"loss": 0.7454,
"step": 1500
},
{
"epoch": 0.12,
"learning_rate": 1.9939354408106307e-05,
"loss": 0.6372,
"step": 2000
},
{
"epoch": 0.15,
"learning_rate": 1.992418542564165e-05,
"loss": 0.5621,
"step": 2500
},
{
"epoch": 0.18,
"learning_rate": 1.9909016443176995e-05,
"loss": 0.5135,
"step": 3000
},
{
"epoch": 0.21,
"learning_rate": 1.9893847460712338e-05,
"loss": 0.4771,
"step": 3500
},
{
"epoch": 0.24,
"learning_rate": 1.987867847824768e-05,
"loss": 0.4487,
"step": 4000
},
{
"epoch": 0.27,
"learning_rate": 1.9863509495783026e-05,
"loss": 0.4204,
"step": 4500
},
{
"epoch": 0.3,
"learning_rate": 1.984834051331837e-05,
"loss": 0.4087,
"step": 5000
},
{
"epoch": 0.33,
"learning_rate": 1.983317153085371e-05,
"loss": 0.3839,
"step": 5500
},
{
"epoch": 0.36,
"learning_rate": 1.9818032886353985e-05,
"loss": 0.3703,
"step": 6000
},
{
"epoch": 0.39,
"learning_rate": 1.980289424185426e-05,
"loss": 0.3623,
"step": 6500
},
{
"epoch": 0.42,
"learning_rate": 1.9787725259389602e-05,
"loss": 0.3517,
"step": 7000
},
{
"epoch": 0.46,
"learning_rate": 1.9772556276924944e-05,
"loss": 0.3371,
"step": 7500
},
{
"epoch": 0.49,
"learning_rate": 1.9757387294460287e-05,
"loss": 0.3307,
"step": 8000
},
{
"epoch": 0.52,
"learning_rate": 1.9742218311995633e-05,
"loss": 0.3152,
"step": 8500
},
{
"epoch": 0.55,
"learning_rate": 1.9727049329530975e-05,
"loss": 0.317,
"step": 9000
},
{
"epoch": 0.58,
"learning_rate": 1.971188034706632e-05,
"loss": 0.3083,
"step": 9500
},
{
"epoch": 0.61,
"learning_rate": 1.9696741702566592e-05,
"loss": 0.3004,
"step": 10000
},
{
"epoch": 0.64,
"learning_rate": 1.9681572720101938e-05,
"loss": 0.2944,
"step": 10500
},
{
"epoch": 0.67,
"learning_rate": 1.966640373763728e-05,
"loss": 0.2949,
"step": 11000
},
{
"epoch": 0.7,
"learning_rate": 1.9651234755172626e-05,
"loss": 0.2863,
"step": 11500
},
{
"epoch": 0.73,
"learning_rate": 1.963606577270797e-05,
"loss": 0.2824,
"step": 12000
},
{
"epoch": 0.76,
"learning_rate": 1.962089679024331e-05,
"loss": 0.279,
"step": 12500
},
{
"epoch": 0.79,
"learning_rate": 1.9605727807778657e-05,
"loss": 0.2754,
"step": 13000
},
{
"epoch": 0.82,
"learning_rate": 1.9590558825314e-05,
"loss": 0.269,
"step": 13500
},
{
"epoch": 0.85,
"learning_rate": 1.9575389842849345e-05,
"loss": 0.2711,
"step": 14000
},
{
"epoch": 0.88,
"learning_rate": 1.9560220860384688e-05,
"loss": 0.2646,
"step": 14500
},
{
"epoch": 0.91,
"learning_rate": 1.9545082215884962e-05,
"loss": 0.2578,
"step": 15000
},
{
"epoch": 0.94,
"learning_rate": 1.9529913233420305e-05,
"loss": 0.2546,
"step": 15500
},
{
"epoch": 0.97,
"learning_rate": 1.9514744250955647e-05,
"loss": 0.2549,
"step": 16000
},
{
"epoch": 1.0,
"eval_bleu": 83.5257,
"eval_gen_len": 14.5799,
"eval_loss": 0.22406013309955597,
"eval_runtime": 225.5638,
"eval_samples_per_second": 71.483,
"eval_steps_per_second": 2.234,
"step": 16481
},
{
"epoch": 1.0,
"learning_rate": 1.949957526849099e-05,
"loss": 0.253,
"step": 16500
},
{
"epoch": 1.03,
"learning_rate": 1.9484406286026335e-05,
"loss": 0.2352,
"step": 17000
},
{
"epoch": 1.06,
"learning_rate": 1.946926764152661e-05,
"loss": 0.2366,
"step": 17500
},
{
"epoch": 1.09,
"learning_rate": 1.9454098659061952e-05,
"loss": 0.2336,
"step": 18000
},
{
"epoch": 1.12,
"learning_rate": 1.9438929676597295e-05,
"loss": 0.2336,
"step": 18500
},
{
"epoch": 1.15,
"learning_rate": 1.9423760694132637e-05,
"loss": 0.2314,
"step": 19000
},
{
"epoch": 1.18,
"learning_rate": 1.940862204963291e-05,
"loss": 0.2258,
"step": 19500
},
{
"epoch": 1.21,
"learning_rate": 1.9393453067168254e-05,
"loss": 0.2252,
"step": 20000
},
{
"epoch": 1.24,
"learning_rate": 1.93782840847036e-05,
"loss": 0.2236,
"step": 20500
},
{
"epoch": 1.27,
"learning_rate": 1.9363115102238942e-05,
"loss": 0.2223,
"step": 21000
},
{
"epoch": 1.3,
"learning_rate": 1.9347946119774288e-05,
"loss": 0.2214,
"step": 21500
},
{
"epoch": 1.33,
"learning_rate": 1.933277713730963e-05,
"loss": 0.218,
"step": 22000
},
{
"epoch": 1.37,
"learning_rate": 1.9317608154844976e-05,
"loss": 0.22,
"step": 22500
},
{
"epoch": 1.4,
"learning_rate": 1.930243917238032e-05,
"loss": 0.2198,
"step": 23000
},
{
"epoch": 1.43,
"learning_rate": 1.928727018991566e-05,
"loss": 0.2183,
"step": 23500
},
{
"epoch": 1.46,
"learning_rate": 1.9272131545415936e-05,
"loss": 0.2176,
"step": 24000
},
{
"epoch": 1.49,
"learning_rate": 1.9256962562951278e-05,
"loss": 0.212,
"step": 24500
},
{
"epoch": 1.52,
"learning_rate": 1.9241823918451552e-05,
"loss": 0.2111,
"step": 25000
},
{
"epoch": 1.55,
"learning_rate": 1.9226654935986895e-05,
"loss": 0.2132,
"step": 25500
},
{
"epoch": 1.58,
"learning_rate": 1.921148595352224e-05,
"loss": 0.2081,
"step": 26000
},
{
"epoch": 1.61,
"learning_rate": 1.9196316971057583e-05,
"loss": 0.2108,
"step": 26500
},
{
"epoch": 1.64,
"learning_rate": 1.918114798859293e-05,
"loss": 0.2078,
"step": 27000
},
{
"epoch": 1.67,
"learning_rate": 1.916597900612827e-05,
"loss": 0.2063,
"step": 27500
},
{
"epoch": 1.7,
"learning_rate": 1.9150810023663614e-05,
"loss": 0.2087,
"step": 28000
},
{
"epoch": 1.73,
"learning_rate": 1.9135641041198957e-05,
"loss": 0.2068,
"step": 28500
},
{
"epoch": 1.76,
"learning_rate": 1.91204720587343e-05,
"loss": 0.2031,
"step": 29000
},
{
"epoch": 1.79,
"learning_rate": 1.9105303076269645e-05,
"loss": 0.2018,
"step": 29500
},
{
"epoch": 1.82,
"learning_rate": 1.9090134093804987e-05,
"loss": 0.2015,
"step": 30000
},
{
"epoch": 1.85,
"learning_rate": 1.9074965111340333e-05,
"loss": 0.2039,
"step": 30500
},
{
"epoch": 1.88,
"learning_rate": 1.9059796128875676e-05,
"loss": 0.2015,
"step": 31000
},
{
"epoch": 1.91,
"learning_rate": 1.904465748437595e-05,
"loss": 0.2013,
"step": 31500
},
{
"epoch": 1.94,
"learning_rate": 1.9029488501911292e-05,
"loss": 0.1983,
"step": 32000
},
{
"epoch": 1.97,
"learning_rate": 1.9014349857411567e-05,
"loss": 0.2012,
"step": 32500
},
{
"epoch": 2.0,
"eval_bleu": 85.8326,
"eval_gen_len": 14.5856,
"eval_loss": 0.18299731612205505,
"eval_runtime": 219.7356,
"eval_samples_per_second": 73.379,
"eval_steps_per_second": 2.294,
"step": 32962
},
{
"epoch": 2.0,
"learning_rate": 1.899918087494691e-05,
"loss": 0.1988,
"step": 33000
},
{
"epoch": 2.03,
"learning_rate": 1.8984011892482255e-05,
"loss": 0.1824,
"step": 33500
},
{
"epoch": 2.06,
"learning_rate": 1.8968873247982526e-05,
"loss": 0.1835,
"step": 34000
},
{
"epoch": 2.09,
"learning_rate": 1.8953704265517872e-05,
"loss": 0.1822,
"step": 34500
},
{
"epoch": 2.12,
"learning_rate": 1.8938535283053214e-05,
"loss": 0.1818,
"step": 35000
},
{
"epoch": 2.15,
"learning_rate": 1.8923366300588557e-05,
"loss": 0.1792,
"step": 35500
},
{
"epoch": 2.18,
"learning_rate": 1.8908197318123903e-05,
"loss": 0.1818,
"step": 36000
},
{
"epoch": 2.21,
"learning_rate": 1.8893058673624173e-05,
"loss": 0.1797,
"step": 36500
},
{
"epoch": 2.25,
"learning_rate": 1.887788969115952e-05,
"loss": 0.178,
"step": 37000
},
{
"epoch": 2.28,
"learning_rate": 1.8862720708694862e-05,
"loss": 0.1795,
"step": 37500
},
{
"epoch": 2.31,
"learning_rate": 1.8847551726230208e-05,
"loss": 0.1804,
"step": 38000
},
{
"epoch": 2.34,
"learning_rate": 1.883238274376555e-05,
"loss": 0.1778,
"step": 38500
},
{
"epoch": 2.37,
"learning_rate": 1.8817213761300893e-05,
"loss": 0.182,
"step": 39000
},
{
"epoch": 2.4,
"learning_rate": 1.880204477883624e-05,
"loss": 0.1773,
"step": 39500
},
{
"epoch": 2.43,
"learning_rate": 1.878687579637158e-05,
"loss": 0.1775,
"step": 40000
},
{
"epoch": 2.46,
"learning_rate": 1.8771706813906924e-05,
"loss": 0.1777,
"step": 40500
},
{
"epoch": 2.49,
"learning_rate": 1.8756537831442266e-05,
"loss": 0.1766,
"step": 41000
},
{
"epoch": 2.52,
"learning_rate": 1.874139918694254e-05,
"loss": 0.177,
"step": 41500
},
{
"epoch": 2.55,
"learning_rate": 1.8726230204477883e-05,
"loss": 0.1757,
"step": 42000
},
{
"epoch": 2.58,
"learning_rate": 1.8711091559978157e-05,
"loss": 0.1761,
"step": 42500
},
{
"epoch": 2.61,
"learning_rate": 1.8695922577513503e-05,
"loss": 0.1747,
"step": 43000
},
{
"epoch": 2.64,
"learning_rate": 1.8680753595048845e-05,
"loss": 0.1756,
"step": 43500
},
{
"epoch": 2.67,
"learning_rate": 1.8665584612584188e-05,
"loss": 0.1775,
"step": 44000
},
{
"epoch": 2.7,
"learning_rate": 1.8650415630119534e-05,
"loss": 0.1745,
"step": 44500
},
{
"epoch": 2.73,
"learning_rate": 1.8635246647654876e-05,
"loss": 0.1721,
"step": 45000
},
{
"epoch": 2.76,
"learning_rate": 1.8620077665190222e-05,
"loss": 0.1742,
"step": 45500
},
{
"epoch": 2.79,
"learning_rate": 1.8604908682725565e-05,
"loss": 0.1721,
"step": 46000
},
{
"epoch": 2.82,
"learning_rate": 1.8589739700260907e-05,
"loss": 0.1734,
"step": 46500
},
{
"epoch": 2.85,
"learning_rate": 1.8574570717796253e-05,
"loss": 0.1729,
"step": 47000
},
{
"epoch": 2.88,
"learning_rate": 1.8559401735331595e-05,
"loss": 0.1719,
"step": 47500
},
{
"epoch": 2.91,
"learning_rate": 1.854423275286694e-05,
"loss": 0.1708,
"step": 48000
},
{
"epoch": 2.94,
"learning_rate": 1.8529094108367212e-05,
"loss": 0.1728,
"step": 48500
},
{
"epoch": 2.97,
"learning_rate": 1.8513925125902558e-05,
"loss": 0.1747,
"step": 49000
},
{
"epoch": 3.0,
"eval_bleu": 86.3628,
"eval_gen_len": 14.5561,
"eval_loss": 0.16949382424354553,
"eval_runtime": 217.2797,
"eval_samples_per_second": 74.208,
"eval_steps_per_second": 2.32,
"step": 49443
},
{
"epoch": 3.0,
"learning_rate": 1.849878648140283e-05,
"loss": 0.1677,
"step": 49500
},
{
"epoch": 3.03,
"learning_rate": 1.8483617498938175e-05,
"loss": 0.1542,
"step": 50000
},
{
"epoch": 3.06,
"learning_rate": 1.8468448516473517e-05,
"loss": 0.1569,
"step": 50500
},
{
"epoch": 3.09,
"learning_rate": 1.845327953400886e-05,
"loss": 0.1591,
"step": 51000
},
{
"epoch": 3.12,
"learning_rate": 1.8438110551544206e-05,
"loss": 0.1543,
"step": 51500
},
{
"epoch": 3.16,
"learning_rate": 1.8422971907044476e-05,
"loss": 0.1557,
"step": 52000
},
{
"epoch": 3.19,
"learning_rate": 1.8407802924579822e-05,
"loss": 0.1572,
"step": 52500
},
{
"epoch": 3.22,
"learning_rate": 1.8392633942115165e-05,
"loss": 0.1582,
"step": 53000
},
{
"epoch": 3.25,
"learning_rate": 1.8377464959650507e-05,
"loss": 0.1608,
"step": 53500
},
{
"epoch": 3.28,
"learning_rate": 1.836229597718585e-05,
"loss": 0.1535,
"step": 54000
},
{
"epoch": 3.31,
"learning_rate": 1.8347126994721196e-05,
"loss": 0.1571,
"step": 54500
},
{
"epoch": 3.34,
"learning_rate": 1.8331958012256538e-05,
"loss": 0.1548,
"step": 55000
},
{
"epoch": 3.37,
"learning_rate": 1.8316789029791884e-05,
"loss": 0.154,
"step": 55500
},
{
"epoch": 3.4,
"learning_rate": 1.8301650385292155e-05,
"loss": 0.1587,
"step": 56000
},
{
"epoch": 3.43,
"learning_rate": 1.82864814028275e-05,
"loss": 0.1552,
"step": 56500
},
{
"epoch": 3.46,
"learning_rate": 1.8271312420362843e-05,
"loss": 0.1499,
"step": 57000
},
{
"epoch": 3.49,
"learning_rate": 1.8256173775863117e-05,
"loss": 0.1564,
"step": 57500
},
{
"epoch": 3.52,
"learning_rate": 1.824100479339846e-05,
"loss": 0.1576,
"step": 58000
},
{
"epoch": 3.55,
"learning_rate": 1.8225835810933802e-05,
"loss": 0.1568,
"step": 58500
},
{
"epoch": 3.58,
"learning_rate": 1.8210666828469148e-05,
"loss": 0.1549,
"step": 59000
},
{
"epoch": 3.61,
"learning_rate": 1.819549784600449e-05,
"loss": 0.1531,
"step": 59500
},
{
"epoch": 3.64,
"learning_rate": 1.8180328863539837e-05,
"loss": 0.1563,
"step": 60000
},
{
"epoch": 3.67,
"learning_rate": 1.816515988107518e-05,
"loss": 0.1555,
"step": 60500
},
{
"epoch": 3.7,
"learning_rate": 1.8149990898610525e-05,
"loss": 0.1539,
"step": 61000
},
{
"epoch": 3.73,
"learning_rate": 1.8134882592075724e-05,
"loss": 0.1519,
"step": 61500
},
{
"epoch": 3.76,
"learning_rate": 1.811971360961107e-05,
"loss": 0.1552,
"step": 62000
},
{
"epoch": 3.79,
"learning_rate": 1.8104544627146412e-05,
"loss": 0.1545,
"step": 62500
},
{
"epoch": 3.82,
"learning_rate": 1.808937564468176e-05,
"loss": 0.1505,
"step": 63000
},
{
"epoch": 3.85,
"learning_rate": 1.80742066622171e-05,
"loss": 0.1551,
"step": 63500
},
{
"epoch": 3.88,
"learning_rate": 1.8059037679752443e-05,
"loss": 0.1553,
"step": 64000
},
{
"epoch": 3.91,
"learning_rate": 1.8043868697287786e-05,
"loss": 0.1524,
"step": 64500
},
{
"epoch": 3.94,
"learning_rate": 1.802873005278806e-05,
"loss": 0.1552,
"step": 65000
},
{
"epoch": 3.97,
"learning_rate": 1.8013561070323406e-05,
"loss": 0.1528,
"step": 65500
},
{
"epoch": 4.0,
"eval_bleu": 86.9553,
"eval_gen_len": 14.6311,
"eval_loss": 0.16092169284820557,
"eval_runtime": 219.7454,
"eval_samples_per_second": 73.376,
"eval_steps_per_second": 2.294,
"step": 65924
},
{
"epoch": 4.0,
"learning_rate": 1.799839208785875e-05,
"loss": 0.1528,
"step": 66000
},
{
"epoch": 4.03,
"learning_rate": 1.798322310539409e-05,
"loss": 0.1406,
"step": 66500
},
{
"epoch": 4.07,
"learning_rate": 1.7968054122929433e-05,
"loss": 0.1436,
"step": 67000
},
{
"epoch": 4.1,
"learning_rate": 1.795288514046478e-05,
"loss": 0.1407,
"step": 67500
},
{
"epoch": 4.13,
"learning_rate": 1.7937716158000122e-05,
"loss": 0.137,
"step": 68000
},
{
"epoch": 4.16,
"learning_rate": 1.7922547175535468e-05,
"loss": 0.1403,
"step": 68500
},
{
"epoch": 4.19,
"learning_rate": 1.790740853103574e-05,
"loss": 0.1405,
"step": 69000
},
{
"epoch": 4.22,
"learning_rate": 1.7892269886536013e-05,
"loss": 0.1408,
"step": 69500
},
{
"epoch": 4.25,
"learning_rate": 1.7877100904071355e-05,
"loss": 0.1354,
"step": 70000
},
{
"epoch": 4.28,
"learning_rate": 1.786196225957163e-05,
"loss": 0.1426,
"step": 70500
},
{
"epoch": 4.31,
"learning_rate": 1.7846793277106972e-05,
"loss": 0.1411,
"step": 71000
},
{
"epoch": 4.34,
"learning_rate": 1.7831624294642318e-05,
"loss": 0.1415,
"step": 71500
},
{
"epoch": 4.37,
"learning_rate": 1.781645531217766e-05,
"loss": 0.1398,
"step": 72000
},
{
"epoch": 4.4,
"learning_rate": 1.7801286329713003e-05,
"loss": 0.1398,
"step": 72500
},
{
"epoch": 4.43,
"learning_rate": 1.778611734724835e-05,
"loss": 0.139,
"step": 73000
},
{
"epoch": 4.46,
"learning_rate": 1.777094836478369e-05,
"loss": 0.142,
"step": 73500
},
{
"epoch": 4.49,
"learning_rate": 1.7755779382319037e-05,
"loss": 0.1423,
"step": 74000
},
{
"epoch": 4.52,
"learning_rate": 1.774061039985438e-05,
"loss": 0.1399,
"step": 74500
},
{
"epoch": 4.55,
"learning_rate": 1.7725441417389725e-05,
"loss": 0.1417,
"step": 75000
},
{
"epoch": 4.58,
"learning_rate": 1.7710302772889996e-05,
"loss": 0.141,
"step": 75500
},
{
"epoch": 4.61,
"learning_rate": 1.7695133790425342e-05,
"loss": 0.1434,
"step": 76000
},
{
"epoch": 4.64,
"learning_rate": 1.7679964807960684e-05,
"loss": 0.1372,
"step": 76500
},
{
"epoch": 4.67,
"learning_rate": 1.7664795825496027e-05,
"loss": 0.1399,
"step": 77000
},
{
"epoch": 4.7,
"learning_rate": 1.764962684303137e-05,
"loss": 0.1399,
"step": 77500
},
{
"epoch": 4.73,
"learning_rate": 1.7634457860566715e-05,
"loss": 0.1388,
"step": 78000
},
{
"epoch": 4.76,
"learning_rate": 1.7619288878102058e-05,
"loss": 0.1399,
"step": 78500
},
{
"epoch": 4.79,
"learning_rate": 1.7604150233602332e-05,
"loss": 0.1402,
"step": 79000
},
{
"epoch": 4.82,
"learning_rate": 1.7588981251137675e-05,
"loss": 0.1379,
"step": 79500
},
{
"epoch": 4.85,
"learning_rate": 1.7573812268673017e-05,
"loss": 0.1392,
"step": 80000
},
{
"epoch": 4.88,
"learning_rate": 1.7558643286208363e-05,
"loss": 0.1379,
"step": 80500
},
{
"epoch": 4.91,
"learning_rate": 1.7543474303743705e-05,
"loss": 0.1392,
"step": 81000
},
{
"epoch": 4.95,
"learning_rate": 1.752830532127905e-05,
"loss": 0.1397,
"step": 81500
},
{
"epoch": 4.98,
"learning_rate": 1.7513166676779322e-05,
"loss": 0.1398,
"step": 82000
},
{
"epoch": 5.0,
"eval_bleu": 87.362,
"eval_gen_len": 14.6795,
"eval_loss": 0.15578030049800873,
"eval_runtime": 221.0197,
"eval_samples_per_second": 72.953,
"eval_steps_per_second": 2.28,
"step": 82405
},
{
"epoch": 5.01,
"learning_rate": 1.7497997694314668e-05,
"loss": 0.1368,
"step": 82500
},
{
"epoch": 5.04,
"learning_rate": 1.748282871185001e-05,
"loss": 0.1257,
"step": 83000
},
{
"epoch": 5.07,
"learning_rate": 1.7467659729385353e-05,
"loss": 0.1253,
"step": 83500
},
{
"epoch": 5.1,
"learning_rate": 1.7452521084885627e-05,
"loss": 0.1295,
"step": 84000
},
{
"epoch": 5.13,
"learning_rate": 1.743735210242097e-05,
"loss": 0.1284,
"step": 84500
},
{
"epoch": 5.16,
"learning_rate": 1.7422183119956316e-05,
"loss": 0.1255,
"step": 85000
},
{
"epoch": 5.19,
"learning_rate": 1.7407014137491658e-05,
"loss": 0.1285,
"step": 85500
},
{
"epoch": 5.22,
"learning_rate": 1.7391845155027004e-05,
"loss": 0.129,
"step": 86000
},
{
"epoch": 5.25,
"learning_rate": 1.7376676172562346e-05,
"loss": 0.1286,
"step": 86500
},
{
"epoch": 5.28,
"learning_rate": 1.736150719009769e-05,
"loss": 0.1287,
"step": 87000
},
{
"epoch": 5.31,
"learning_rate": 1.7346368545597963e-05,
"loss": 0.1313,
"step": 87500
},
{
"epoch": 5.34,
"learning_rate": 1.733119956313331e-05,
"loss": 0.1312,
"step": 88000
},
{
"epoch": 5.37,
"learning_rate": 1.731603058066865e-05,
"loss": 0.129,
"step": 88500
},
{
"epoch": 5.4,
"learning_rate": 1.7300861598203994e-05,
"loss": 0.1262,
"step": 89000
},
{
"epoch": 5.43,
"learning_rate": 1.7285692615739336e-05,
"loss": 0.128,
"step": 89500
},
{
"epoch": 5.46,
"learning_rate": 1.727052363327468e-05,
"loss": 0.1298,
"step": 90000
},
{
"epoch": 5.49,
"learning_rate": 1.7255354650810025e-05,
"loss": 0.1299,
"step": 90500
},
{
"epoch": 5.52,
"learning_rate": 1.7240185668345367e-05,
"loss": 0.1291,
"step": 91000
},
{
"epoch": 5.55,
"learning_rate": 1.7225016685880713e-05,
"loss": 0.1287,
"step": 91500
},
{
"epoch": 5.58,
"learning_rate": 1.7209847703416056e-05,
"loss": 0.1252,
"step": 92000
},
{
"epoch": 5.61,
"learning_rate": 1.71946787209514e-05,
"loss": 0.1277,
"step": 92500
},
{
"epoch": 5.64,
"learning_rate": 1.7179509738486744e-05,
"loss": 0.1244,
"step": 93000
},
{
"epoch": 5.67,
"learning_rate": 1.7164401431951947e-05,
"loss": 0.1278,
"step": 93500
},
{
"epoch": 5.7,
"learning_rate": 1.714923244948729e-05,
"loss": 0.1299,
"step": 94000
},
{
"epoch": 5.73,
"learning_rate": 1.713406346702263e-05,
"loss": 0.1284,
"step": 94500
},
{
"epoch": 5.76,
"learning_rate": 1.7118894484557977e-05,
"loss": 0.1273,
"step": 95000
},
{
"epoch": 5.79,
"learning_rate": 1.7103755840058248e-05,
"loss": 0.1281,
"step": 95500
},
{
"epoch": 5.82,
"learning_rate": 1.7088586857593594e-05,
"loss": 0.1287,
"step": 96000
},
{
"epoch": 5.86,
"learning_rate": 1.7073417875128937e-05,
"loss": 0.1273,
"step": 96500
},
{
"epoch": 5.89,
"learning_rate": 1.7058248892664283e-05,
"loss": 0.1291,
"step": 97000
},
{
"epoch": 5.92,
"learning_rate": 1.7043079910199625e-05,
"loss": 0.1281,
"step": 97500
},
{
"epoch": 5.95,
"learning_rate": 1.702791092773497e-05,
"loss": 0.1251,
"step": 98000
},
{
"epoch": 5.98,
"learning_rate": 1.7012772283235242e-05,
"loss": 0.1252,
"step": 98500
},
{
"epoch": 6.0,
"eval_bleu": 87.4518,
"eval_gen_len": 14.6618,
"eval_loss": 0.1534184217453003,
"eval_runtime": 219.098,
"eval_samples_per_second": 73.593,
"eval_steps_per_second": 2.3,
"step": 98886
},
{
"epoch": 6.01,
"learning_rate": 1.6997603300770588e-05,
"loss": 0.122,
"step": 99000
},
{
"epoch": 6.04,
"learning_rate": 1.698243431830593e-05,
"loss": 0.1178,
"step": 99500
},
{
"epoch": 6.07,
"learning_rate": 1.6967265335841273e-05,
"loss": 0.1152,
"step": 100000
},
{
"epoch": 6.1,
"learning_rate": 1.6952126691341547e-05,
"loss": 0.117,
"step": 100500
},
{
"epoch": 6.13,
"learning_rate": 1.693695770887689e-05,
"loss": 0.1158,
"step": 101000
},
{
"epoch": 6.16,
"learning_rate": 1.6921788726412235e-05,
"loss": 0.1169,
"step": 101500
},
{
"epoch": 6.19,
"learning_rate": 1.6906619743947578e-05,
"loss": 0.1164,
"step": 102000
},
{
"epoch": 6.22,
"learning_rate": 1.689145076148292e-05,
"loss": 0.1191,
"step": 102500
},
{
"epoch": 6.25,
"learning_rate": 1.6876281779018263e-05,
"loss": 0.1186,
"step": 103000
},
{
"epoch": 6.28,
"learning_rate": 1.686111279655361e-05,
"loss": 0.1184,
"step": 103500
},
{
"epoch": 6.31,
"learning_rate": 1.684594381408895e-05,
"loss": 0.1162,
"step": 104000
},
{
"epoch": 6.34,
"learning_rate": 1.6830774831624297e-05,
"loss": 0.1172,
"step": 104500
},
{
"epoch": 6.37,
"learning_rate": 1.681560584915964e-05,
"loss": 0.1188,
"step": 105000
},
{
"epoch": 6.4,
"learning_rate": 1.6800467204659914e-05,
"loss": 0.1174,
"step": 105500
},
{
"epoch": 6.43,
"learning_rate": 1.6785298222195256e-05,
"loss": 0.1194,
"step": 106000
},
{
"epoch": 6.46,
"learning_rate": 1.67701292397306e-05,
"loss": 0.1187,
"step": 106500
},
{
"epoch": 6.49,
"learning_rate": 1.6754960257265944e-05,
"loss": 0.1172,
"step": 107000
},
{
"epoch": 6.52,
"learning_rate": 1.6739821612766215e-05,
"loss": 0.1185,
"step": 107500
},
{
"epoch": 6.55,
"learning_rate": 1.672465263030156e-05,
"loss": 0.1184,
"step": 108000
},
{
"epoch": 6.58,
"learning_rate": 1.6709483647836904e-05,
"loss": 0.1182,
"step": 108500
},
{
"epoch": 6.61,
"learning_rate": 1.669431466537225e-05,
"loss": 0.1174,
"step": 109000
},
{
"epoch": 6.64,
"learning_rate": 1.6679145682907592e-05,
"loss": 0.1183,
"step": 109500
},
{
"epoch": 6.67,
"learning_rate": 1.6663976700442938e-05,
"loss": 0.1186,
"step": 110000
},
{
"epoch": 6.7,
"learning_rate": 1.664880771797828e-05,
"loss": 0.1161,
"step": 110500
},
{
"epoch": 6.74,
"learning_rate": 1.6633638735513623e-05,
"loss": 0.1186,
"step": 111000
},
{
"epoch": 6.77,
"learning_rate": 1.6618469753048965e-05,
"loss": 0.1172,
"step": 111500
},
{
"epoch": 6.8,
"learning_rate": 1.660333110854924e-05,
"loss": 0.118,
"step": 112000
},
{
"epoch": 6.83,
"learning_rate": 1.6588162126084585e-05,
"loss": 0.1171,
"step": 112500
},
{
"epoch": 6.86,
"learning_rate": 1.6572993143619928e-05,
"loss": 0.1166,
"step": 113000
},
{
"epoch": 6.89,
"learning_rate": 1.655782416115527e-05,
"loss": 0.1189,
"step": 113500
},
{
"epoch": 6.92,
"learning_rate": 1.6542685516655545e-05,
"loss": 0.1186,
"step": 114000
},
{
"epoch": 6.95,
"learning_rate": 1.6527516534190887e-05,
"loss": 0.1187,
"step": 114500
},
{
"epoch": 6.98,
"learning_rate": 1.651234755172623e-05,
"loss": 0.1173,
"step": 115000
},
{
"epoch": 7.0,
"eval_bleu": 87.3802,
"eval_gen_len": 14.6933,
"eval_loss": 0.15458865463733673,
"eval_runtime": 219.3387,
"eval_samples_per_second": 73.512,
"eval_steps_per_second": 2.298,
"step": 115367
},
{
"epoch": 7.01,
"learning_rate": 1.6497178569261575e-05,
"loss": 0.1161,
"step": 115500
},
{
"epoch": 7.04,
"learning_rate": 1.6482009586796918e-05,
"loss": 0.1073,
"step": 116000
},
{
"epoch": 7.07,
"learning_rate": 1.6466840604332264e-05,
"loss": 0.1058,
"step": 116500
},
{
"epoch": 7.1,
"learning_rate": 1.6451671621867606e-05,
"loss": 0.1068,
"step": 117000
},
{
"epoch": 7.13,
"learning_rate": 1.643653297736788e-05,
"loss": 0.1093,
"step": 117500
},
{
"epoch": 7.16,
"learning_rate": 1.6421363994903223e-05,
"loss": 0.1076,
"step": 118000
},
{
"epoch": 7.19,
"learning_rate": 1.6406195012438566e-05,
"loss": 0.1089,
"step": 118500
},
{
"epoch": 7.22,
"learning_rate": 1.639102602997391e-05,
"loss": 0.1094,
"step": 119000
},
{
"epoch": 7.25,
"learning_rate": 1.6375857047509254e-05,
"loss": 0.1074,
"step": 119500
},
{
"epoch": 7.28,
"learning_rate": 1.63606880650446e-05,
"loss": 0.1098,
"step": 120000
},
{
"epoch": 7.31,
"learning_rate": 1.6345519082579942e-05,
"loss": 0.1083,
"step": 120500
},
{
"epoch": 7.34,
"learning_rate": 1.6330380438080216e-05,
"loss": 0.1083,
"step": 121000
},
{
"epoch": 7.37,
"learning_rate": 1.631521145561556e-05,
"loss": 0.1098,
"step": 121500
},
{
"epoch": 7.4,
"learning_rate": 1.6300042473150905e-05,
"loss": 0.1066,
"step": 122000
},
{
"epoch": 7.43,
"learning_rate": 1.6284873490686247e-05,
"loss": 0.1075,
"step": 122500
},
{
"epoch": 7.46,
"learning_rate": 1.626970450822159e-05,
"loss": 0.1086,
"step": 123000
},
{
"epoch": 7.49,
"learning_rate": 1.6254565863721864e-05,
"loss": 0.1078,
"step": 123500
},
{
"epoch": 7.52,
"learning_rate": 1.6239396881257207e-05,
"loss": 0.1119,
"step": 124000
},
{
"epoch": 7.55,
"learning_rate": 1.622422789879255e-05,
"loss": 0.1079,
"step": 124500
},
{
"epoch": 7.58,
"learning_rate": 1.6209058916327895e-05,
"loss": 0.1086,
"step": 125000
},
{
"epoch": 7.61,
"learning_rate": 1.6193889933863237e-05,
"loss": 0.1091,
"step": 125500
},
{
"epoch": 7.65,
"learning_rate": 1.617875128936351e-05,
"loss": 0.1075,
"step": 126000
},
{
"epoch": 7.68,
"learning_rate": 1.6163582306898854e-05,
"loss": 0.1102,
"step": 126500
},
{
"epoch": 7.71,
"learning_rate": 1.6148413324434197e-05,
"loss": 0.1097,
"step": 127000
},
{
"epoch": 7.74,
"learning_rate": 1.6133244341969542e-05,
"loss": 0.1077,
"step": 127500
},
{
"epoch": 7.77,
"learning_rate": 1.6118075359504885e-05,
"loss": 0.1106,
"step": 128000
},
{
"epoch": 7.8,
"learning_rate": 1.610290637704023e-05,
"loss": 0.1095,
"step": 128500
},
{
"epoch": 7.83,
"learning_rate": 1.6087737394575573e-05,
"loss": 0.1104,
"step": 129000
},
{
"epoch": 7.86,
"learning_rate": 1.6072598750075844e-05,
"loss": 0.1097,
"step": 129500
},
{
"epoch": 7.89,
"learning_rate": 1.605742976761119e-05,
"loss": 0.1085,
"step": 130000
},
{
"epoch": 7.92,
"learning_rate": 1.6042260785146533e-05,
"loss": 0.1091,
"step": 130500
},
{
"epoch": 7.95,
"learning_rate": 1.602709180268188e-05,
"loss": 0.1088,
"step": 131000
},
{
"epoch": 7.98,
"learning_rate": 1.601192282021722e-05,
"loss": 0.1092,
"step": 131500
},
{
"epoch": 8.0,
"eval_bleu": 87.7655,
"eval_gen_len": 14.7622,
"eval_loss": 0.15202689170837402,
"eval_runtime": 220.0905,
"eval_samples_per_second": 73.261,
"eval_steps_per_second": 2.29,
"step": 131848
},
{
"epoch": 8.01,
"learning_rate": 1.5996784175717495e-05,
"loss": 0.1078,
"step": 132000
},
{
"epoch": 8.04,
"learning_rate": 1.5981615193252838e-05,
"loss": 0.0986,
"step": 132500
},
{
"epoch": 8.07,
"learning_rate": 1.5966446210788183e-05,
"loss": 0.0993,
"step": 133000
},
{
"epoch": 8.1,
"learning_rate": 1.5951277228323526e-05,
"loss": 0.097,
"step": 133500
},
{
"epoch": 8.13,
"learning_rate": 1.593610824585887e-05,
"loss": 0.0987,
"step": 134000
},
{
"epoch": 8.16,
"learning_rate": 1.5920969601359143e-05,
"loss": 0.0996,
"step": 134500
},
{
"epoch": 8.19,
"learning_rate": 1.590580061889449e-05,
"loss": 0.0987,
"step": 135000
},
{
"epoch": 8.22,
"learning_rate": 1.589063163642983e-05,
"loss": 0.0987,
"step": 135500
},
{
"epoch": 8.25,
"learning_rate": 1.5875462653965174e-05,
"loss": 0.1005,
"step": 136000
},
{
"epoch": 8.28,
"learning_rate": 1.5860293671500516e-05,
"loss": 0.0987,
"step": 136500
},
{
"epoch": 8.31,
"learning_rate": 1.584512468903586e-05,
"loss": 0.1009,
"step": 137000
},
{
"epoch": 8.34,
"learning_rate": 1.5829986044536133e-05,
"loss": 0.1018,
"step": 137500
},
{
"epoch": 8.37,
"learning_rate": 1.581481706207148e-05,
"loss": 0.0992,
"step": 138000
},
{
"epoch": 8.4,
"learning_rate": 1.579964807960682e-05,
"loss": 0.1013,
"step": 138500
},
{
"epoch": 8.43,
"learning_rate": 1.5784479097142164e-05,
"loss": 0.0987,
"step": 139000
},
{
"epoch": 8.46,
"learning_rate": 1.5769340452642438e-05,
"loss": 0.1006,
"step": 139500
},
{
"epoch": 8.49,
"learning_rate": 1.575417147017778e-05,
"loss": 0.1,
"step": 140000
},
{
"epoch": 8.52,
"learning_rate": 1.5739002487713126e-05,
"loss": 0.1023,
"step": 140500
},
{
"epoch": 8.56,
"learning_rate": 1.572383350524847e-05,
"loss": 0.1011,
"step": 141000
},
{
"epoch": 8.59,
"learning_rate": 1.570866452278381e-05,
"loss": 0.1019,
"step": 141500
},
{
"epoch": 8.62,
"learning_rate": 1.5693525878284085e-05,
"loss": 0.0995,
"step": 142000
},
{
"epoch": 8.65,
"learning_rate": 1.567838723378436e-05,
"loss": 0.1015,
"step": 142500
},
{
"epoch": 8.68,
"learning_rate": 1.5663218251319702e-05,
"loss": 0.1,
"step": 143000
},
{
"epoch": 8.71,
"learning_rate": 1.5648049268855045e-05,
"loss": 0.1012,
"step": 143500
},
{
"epoch": 8.74,
"learning_rate": 1.563288028639039e-05,
"loss": 0.1012,
"step": 144000
},
{
"epoch": 8.77,
"learning_rate": 1.5617711303925733e-05,
"loss": 0.103,
"step": 144500
},
{
"epoch": 8.8,
"learning_rate": 1.560254232146108e-05,
"loss": 0.1021,
"step": 145000
},
{
"epoch": 8.83,
"learning_rate": 1.558737333899642e-05,
"loss": 0.103,
"step": 145500
},
{
"epoch": 8.86,
"learning_rate": 1.5572204356531767e-05,
"loss": 0.1016,
"step": 146000
},
{
"epoch": 8.89,
"learning_rate": 1.555703537406711e-05,
"loss": 0.1013,
"step": 146500
},
{
"epoch": 8.92,
"learning_rate": 1.5541866391602452e-05,
"loss": 0.1009,
"step": 147000
},
{
"epoch": 8.95,
"learning_rate": 1.5526727747102726e-05,
"loss": 0.1024,
"step": 147500
},
{
"epoch": 8.98,
"learning_rate": 1.551155876463807e-05,
"loss": 0.1019,
"step": 148000
},
{
"epoch": 9.0,
"eval_bleu": 87.7707,
"eval_gen_len": 14.676,
"eval_loss": 0.15273568034172058,
"eval_runtime": 220.9435,
"eval_samples_per_second": 72.978,
"eval_steps_per_second": 2.281,
"step": 148329
},
{
"epoch": 9.01,
"learning_rate": 1.5496389782173415e-05,
"loss": 0.0979,
"step": 148500
},
{
"epoch": 9.04,
"learning_rate": 1.5481220799708757e-05,
"loss": 0.0906,
"step": 149000
},
{
"epoch": 9.07,
"learning_rate": 1.54660518172441e-05,
"loss": 0.0902,
"step": 149500
},
{
"epoch": 9.1,
"learning_rate": 1.5450882834779442e-05,
"loss": 0.093,
"step": 150000
},
{
"epoch": 9.13,
"learning_rate": 1.5435744190279716e-05,
"loss": 0.0926,
"step": 150500
},
{
"epoch": 9.16,
"learning_rate": 1.542057520781506e-05,
"loss": 0.0905,
"step": 151000
},
{
"epoch": 9.19,
"learning_rate": 1.5405406225350405e-05,
"loss": 0.0915,
"step": 151500
},
{
"epoch": 9.22,
"learning_rate": 1.5390237242885747e-05,
"loss": 0.092,
"step": 152000
},
{
"epoch": 9.25,
"learning_rate": 1.5375068260421093e-05,
"loss": 0.0922,
"step": 152500
},
{
"epoch": 9.28,
"learning_rate": 1.5359899277956436e-05,
"loss": 0.0913,
"step": 153000
},
{
"epoch": 9.31,
"learning_rate": 1.5344730295491778e-05,
"loss": 0.0932,
"step": 153500
},
{
"epoch": 9.34,
"learning_rate": 1.5329591650992052e-05,
"loss": 0.0925,
"step": 154000
},
{
"epoch": 9.37,
"learning_rate": 1.5314422668527395e-05,
"loss": 0.0927,
"step": 154500
},
{
"epoch": 9.4,
"learning_rate": 1.529925368606274e-05,
"loss": 0.0933,
"step": 155000
},
{
"epoch": 9.44,
"learning_rate": 1.5284084703598083e-05,
"loss": 0.0928,
"step": 155500
},
{
"epoch": 9.47,
"learning_rate": 1.5268946059098357e-05,
"loss": 0.094,
"step": 156000
},
{
"epoch": 9.5,
"learning_rate": 1.52537770766337e-05,
"loss": 0.0925,
"step": 156500
},
{
"epoch": 9.53,
"learning_rate": 1.5238608094169046e-05,
"loss": 0.093,
"step": 157000
},
{
"epoch": 9.56,
"learning_rate": 1.5223439111704388e-05,
"loss": 0.0936,
"step": 157500
},
{
"epoch": 9.59,
"learning_rate": 1.5208270129239732e-05,
"loss": 0.0935,
"step": 158000
},
{
"epoch": 9.62,
"learning_rate": 1.5193101146775075e-05,
"loss": 0.0913,
"step": 158500
},
{
"epoch": 9.65,
"learning_rate": 1.5177932164310417e-05,
"loss": 0.0943,
"step": 159000
},
{
"epoch": 9.68,
"learning_rate": 1.5162763181845763e-05,
"loss": 0.093,
"step": 159500
},
{
"epoch": 9.71,
"learning_rate": 1.5147624537346036e-05,
"loss": 0.0943,
"step": 160000
},
{
"epoch": 9.74,
"learning_rate": 1.513248589284631e-05,
"loss": 0.0952,
"step": 160500
},
{
"epoch": 9.77,
"learning_rate": 1.5117316910381653e-05,
"loss": 0.0926,
"step": 161000
},
{
"epoch": 9.8,
"learning_rate": 1.5102147927916997e-05,
"loss": 0.0939,
"step": 161500
},
{
"epoch": 9.83,
"learning_rate": 1.508697894545234e-05,
"loss": 0.0964,
"step": 162000
},
{
"epoch": 9.86,
"learning_rate": 1.5071809962987685e-05,
"loss": 0.0959,
"step": 162500
},
{
"epoch": 9.89,
"learning_rate": 1.5056640980523028e-05,
"loss": 0.0936,
"step": 163000
},
{
"epoch": 9.92,
"learning_rate": 1.5041502336023302e-05,
"loss": 0.0937,
"step": 163500
},
{
"epoch": 9.95,
"learning_rate": 1.5026333353558644e-05,
"loss": 0.094,
"step": 164000
},
{
"epoch": 9.98,
"learning_rate": 1.5011164371093988e-05,
"loss": 0.0965,
"step": 164500
},
{
"epoch": 10.0,
"eval_bleu": 87.6567,
"eval_gen_len": 14.6305,
"eval_loss": 0.1554132103919983,
"eval_runtime": 219.4742,
"eval_samples_per_second": 73.467,
"eval_steps_per_second": 2.296,
"step": 164810
},
{
"epoch": 10.01,
"learning_rate": 1.4995995388629333e-05,
"loss": 0.0908,
"step": 165000
},
{
"epoch": 10.04,
"learning_rate": 1.4980826406164675e-05,
"loss": 0.0832,
"step": 165500
},
{
"epoch": 10.07,
"learning_rate": 1.496568776166495e-05,
"loss": 0.0833,
"step": 166000
},
{
"epoch": 10.1,
"learning_rate": 1.4950518779200292e-05,
"loss": 0.0843,
"step": 166500
},
{
"epoch": 10.13,
"learning_rate": 1.4935349796735636e-05,
"loss": 0.0844,
"step": 167000
},
{
"epoch": 10.16,
"learning_rate": 1.4920180814270979e-05,
"loss": 0.0831,
"step": 167500
},
{
"epoch": 10.19,
"learning_rate": 1.4905011831806324e-05,
"loss": 0.0843,
"step": 168000
},
{
"epoch": 10.22,
"learning_rate": 1.4889842849341667e-05,
"loss": 0.0856,
"step": 168500
},
{
"epoch": 10.25,
"learning_rate": 1.4874673866877011e-05,
"loss": 0.0856,
"step": 169000
},
{
"epoch": 10.28,
"learning_rate": 1.4859504884412355e-05,
"loss": 0.0867,
"step": 169500
},
{
"epoch": 10.31,
"learning_rate": 1.4844396577877558e-05,
"loss": 0.0856,
"step": 170000
},
{
"epoch": 10.35,
"learning_rate": 1.48292275954129e-05,
"loss": 0.0866,
"step": 170500
},
{
"epoch": 10.38,
"learning_rate": 1.4814058612948246e-05,
"loss": 0.0857,
"step": 171000
},
{
"epoch": 10.41,
"learning_rate": 1.4798889630483589e-05,
"loss": 0.0862,
"step": 171500
},
{
"epoch": 10.44,
"learning_rate": 1.4783720648018931e-05,
"loss": 0.0866,
"step": 172000
},
{
"epoch": 10.47,
"learning_rate": 1.4768551665554275e-05,
"loss": 0.0874,
"step": 172500
},
{
"epoch": 10.5,
"learning_rate": 1.4753413021054548e-05,
"loss": 0.0869,
"step": 173000
},
{
"epoch": 10.53,
"learning_rate": 1.4738244038589892e-05,
"loss": 0.0876,
"step": 173500
},
{
"epoch": 10.56,
"learning_rate": 1.4723075056125236e-05,
"loss": 0.0891,
"step": 174000
},
{
"epoch": 10.59,
"learning_rate": 1.470790607366058e-05,
"loss": 0.0866,
"step": 174500
},
{
"epoch": 10.62,
"learning_rate": 1.4692737091195923e-05,
"loss": 0.088,
"step": 175000
},
{
"epoch": 10.65,
"learning_rate": 1.4677598446696197e-05,
"loss": 0.0868,
"step": 175500
},
{
"epoch": 10.68,
"learning_rate": 1.466245980219647e-05,
"loss": 0.0873,
"step": 176000
},
{
"epoch": 10.71,
"learning_rate": 1.4647290819731814e-05,
"loss": 0.0836,
"step": 176500
},
{
"epoch": 10.74,
"learning_rate": 1.4632121837267156e-05,
"loss": 0.0886,
"step": 177000
},
{
"epoch": 10.77,
"learning_rate": 1.4616952854802502e-05,
"loss": 0.0858,
"step": 177500
},
{
"epoch": 10.8,
"learning_rate": 1.4601814210302773e-05,
"loss": 0.0876,
"step": 178000
},
{
"epoch": 10.83,
"learning_rate": 1.4586645227838119e-05,
"loss": 0.0869,
"step": 178500
},
{
"epoch": 10.86,
"learning_rate": 1.4571476245373461e-05,
"loss": 0.0865,
"step": 179000
},
{
"epoch": 10.89,
"learning_rate": 1.4556307262908804e-05,
"loss": 0.0877,
"step": 179500
},
{
"epoch": 10.92,
"learning_rate": 1.454113828044415e-05,
"loss": 0.089,
"step": 180000
},
{
"epoch": 10.95,
"learning_rate": 1.4525969297979492e-05,
"loss": 0.0883,
"step": 180500
},
{
"epoch": 10.98,
"learning_rate": 1.4510800315514836e-05,
"loss": 0.0875,
"step": 181000
},
{
"epoch": 11.0,
"eval_bleu": 87.8907,
"eval_gen_len": 14.7111,
"eval_loss": 0.15501651167869568,
"eval_runtime": 220.1923,
"eval_samples_per_second": 73.227,
"eval_steps_per_second": 2.289,
"step": 181291
},
{
"epoch": 11.01,
"learning_rate": 1.4495631333050179e-05,
"loss": 0.0855,
"step": 181500
},
{
"epoch": 11.04,
"learning_rate": 1.4480462350585525e-05,
"loss": 0.0761,
"step": 182000
},
{
"epoch": 11.07,
"learning_rate": 1.4465293368120867e-05,
"loss": 0.0785,
"step": 182500
},
{
"epoch": 11.1,
"learning_rate": 1.4450124385656213e-05,
"loss": 0.0783,
"step": 183000
},
{
"epoch": 11.13,
"learning_rate": 1.4434985741156484e-05,
"loss": 0.0792,
"step": 183500
},
{
"epoch": 11.16,
"learning_rate": 1.441981675869183e-05,
"loss": 0.0777,
"step": 184000
},
{
"epoch": 11.19,
"learning_rate": 1.4404647776227172e-05,
"loss": 0.0782,
"step": 184500
},
{
"epoch": 11.23,
"learning_rate": 1.4389478793762515e-05,
"loss": 0.0771,
"step": 185000
},
{
"epoch": 11.26,
"learning_rate": 1.4374309811297859e-05,
"loss": 0.0802,
"step": 185500
},
{
"epoch": 11.29,
"learning_rate": 1.4359140828833203e-05,
"loss": 0.079,
"step": 186000
},
{
"epoch": 11.32,
"learning_rate": 1.4343971846368547e-05,
"loss": 0.0791,
"step": 186500
},
{
"epoch": 11.35,
"learning_rate": 1.432883320186882e-05,
"loss": 0.0788,
"step": 187000
},
{
"epoch": 11.38,
"learning_rate": 1.4313694557369094e-05,
"loss": 0.0784,
"step": 187500
},
{
"epoch": 11.41,
"learning_rate": 1.4298525574904437e-05,
"loss": 0.0806,
"step": 188000
},
{
"epoch": 11.44,
"learning_rate": 1.428335659243978e-05,
"loss": 0.0805,
"step": 188500
},
{
"epoch": 11.47,
"learning_rate": 1.4268187609975123e-05,
"loss": 0.0805,
"step": 189000
},
{
"epoch": 11.5,
"learning_rate": 1.4253018627510469e-05,
"loss": 0.0807,
"step": 189500
},
{
"epoch": 11.53,
"learning_rate": 1.4237849645045812e-05,
"loss": 0.0814,
"step": 190000
},
{
"epoch": 11.56,
"learning_rate": 1.4222680662581154e-05,
"loss": 0.079,
"step": 190500
},
{
"epoch": 11.59,
"learning_rate": 1.4207542018081428e-05,
"loss": 0.0796,
"step": 191000
},
{
"epoch": 11.62,
"learning_rate": 1.419237303561677e-05,
"loss": 0.0796,
"step": 191500
},
{
"epoch": 11.65,
"learning_rate": 1.4177204053152117e-05,
"loss": 0.0789,
"step": 192000
},
{
"epoch": 11.68,
"learning_rate": 1.416203507068746e-05,
"loss": 0.0836,
"step": 192500
},
{
"epoch": 11.71,
"learning_rate": 1.4146866088222803e-05,
"loss": 0.0803,
"step": 193000
},
{
"epoch": 11.74,
"learning_rate": 1.4131697105758146e-05,
"loss": 0.08,
"step": 193500
},
{
"epoch": 11.77,
"learning_rate": 1.4116528123293492e-05,
"loss": 0.0828,
"step": 194000
},
{
"epoch": 11.8,
"learning_rate": 1.4101359140828834e-05,
"loss": 0.0817,
"step": 194500
},
{
"epoch": 11.83,
"learning_rate": 1.4086190158364178e-05,
"loss": 0.0807,
"step": 195000
},
{
"epoch": 11.86,
"learning_rate": 1.4071021175899523e-05,
"loss": 0.0815,
"step": 195500
},
{
"epoch": 11.89,
"learning_rate": 1.4055852193434865e-05,
"loss": 0.08,
"step": 196000
},
{
"epoch": 11.92,
"learning_rate": 1.404068321097021e-05,
"loss": 0.0815,
"step": 196500
},
{
"epoch": 11.95,
"learning_rate": 1.4025544566470482e-05,
"loss": 0.0817,
"step": 197000
},
{
"epoch": 11.98,
"learning_rate": 1.4010375584005826e-05,
"loss": 0.0822,
"step": 197500
},
{
"epoch": 12.0,
"eval_bleu": 87.654,
"eval_gen_len": 14.6811,
"eval_loss": 0.15846888720989227,
"eval_runtime": 218.8023,
"eval_samples_per_second": 73.692,
"eval_steps_per_second": 2.303,
"step": 197772
},
{
"epoch": 12.01,
"learning_rate": 1.3995236939506098e-05,
"loss": 0.0786,
"step": 198000
},
{
"epoch": 12.04,
"learning_rate": 1.3980067957041443e-05,
"loss": 0.0711,
"step": 198500
},
{
"epoch": 12.07,
"learning_rate": 1.3964898974576785e-05,
"loss": 0.0716,
"step": 199000
},
{
"epoch": 12.1,
"learning_rate": 1.3949729992112131e-05,
"loss": 0.0719,
"step": 199500
},
{
"epoch": 12.14,
"learning_rate": 1.3934591347612404e-05,
"loss": 0.0736,
"step": 200000
},
{
"epoch": 12.17,
"learning_rate": 1.3919422365147748e-05,
"loss": 0.0722,
"step": 200500
},
{
"epoch": 12.2,
"learning_rate": 1.390425338268309e-05,
"loss": 0.0746,
"step": 201000
},
{
"epoch": 12.23,
"learning_rate": 1.3889084400218436e-05,
"loss": 0.0748,
"step": 201500
},
{
"epoch": 12.26,
"learning_rate": 1.3873945755718707e-05,
"loss": 0.0723,
"step": 202000
},
{
"epoch": 12.29,
"learning_rate": 1.3858776773254053e-05,
"loss": 0.073,
"step": 202500
},
{
"epoch": 12.32,
"learning_rate": 1.3843607790789395e-05,
"loss": 0.0722,
"step": 203000
},
{
"epoch": 12.35,
"learning_rate": 1.3828438808324738e-05,
"loss": 0.0731,
"step": 203500
},
{
"epoch": 12.38,
"learning_rate": 1.3813269825860082e-05,
"loss": 0.0726,
"step": 204000
},
{
"epoch": 12.41,
"learning_rate": 1.3798100843395426e-05,
"loss": 0.0761,
"step": 204500
},
{
"epoch": 12.44,
"learning_rate": 1.378293186093077e-05,
"loss": 0.0744,
"step": 205000
},
{
"epoch": 12.47,
"learning_rate": 1.3767762878466113e-05,
"loss": 0.074,
"step": 205500
},
{
"epoch": 12.5,
"learning_rate": 1.3752593896001459e-05,
"loss": 0.0741,
"step": 206000
},
{
"epoch": 12.53,
"learning_rate": 1.3737424913536801e-05,
"loss": 0.0734,
"step": 206500
},
{
"epoch": 12.56,
"learning_rate": 1.3722255931072144e-05,
"loss": 0.0745,
"step": 207000
},
{
"epoch": 12.59,
"learning_rate": 1.3707086948607488e-05,
"loss": 0.0745,
"step": 207500
},
{
"epoch": 12.62,
"learning_rate": 1.369194830410776e-05,
"loss": 0.0745,
"step": 208000
},
{
"epoch": 12.65,
"learning_rate": 1.3676779321643106e-05,
"loss": 0.0745,
"step": 208500
},
{
"epoch": 12.68,
"learning_rate": 1.3661610339178449e-05,
"loss": 0.0747,
"step": 209000
},
{
"epoch": 12.71,
"learning_rate": 1.3646441356713793e-05,
"loss": 0.0748,
"step": 209500
},
{
"epoch": 12.74,
"learning_rate": 1.3631272374249135e-05,
"loss": 0.0755,
"step": 210000
},
{
"epoch": 12.77,
"learning_rate": 1.3616103391784481e-05,
"loss": 0.0735,
"step": 210500
},
{
"epoch": 12.8,
"learning_rate": 1.3600934409319824e-05,
"loss": 0.0738,
"step": 211000
},
{
"epoch": 12.83,
"learning_rate": 1.3585765426855168e-05,
"loss": 0.0748,
"step": 211500
},
{
"epoch": 12.86,
"learning_rate": 1.357062678235544e-05,
"loss": 0.0752,
"step": 212000
},
{
"epoch": 12.89,
"learning_rate": 1.3555457799890785e-05,
"loss": 0.0749,
"step": 212500
},
{
"epoch": 12.92,
"learning_rate": 1.3540288817426129e-05,
"loss": 0.0745,
"step": 213000
},
{
"epoch": 12.95,
"learning_rate": 1.3525119834961471e-05,
"loss": 0.0752,
"step": 213500
},
{
"epoch": 12.98,
"learning_rate": 1.3509981190461746e-05,
"loss": 0.077,
"step": 214000
},
{
"epoch": 13.0,
"eval_bleu": 87.6388,
"eval_gen_len": 14.7814,
"eval_loss": 0.16246692836284637,
"eval_runtime": 220.1042,
"eval_samples_per_second": 73.256,
"eval_steps_per_second": 2.29,
"step": 214253
},
{
"epoch": 13.01,
"learning_rate": 1.3494812207997088e-05,
"loss": 0.0717,
"step": 214500
},
{
"epoch": 13.05,
"learning_rate": 1.3479643225532432e-05,
"loss": 0.0665,
"step": 215000
},
{
"epoch": 13.08,
"learning_rate": 1.3464474243067775e-05,
"loss": 0.066,
"step": 215500
},
{
"epoch": 13.11,
"learning_rate": 1.3449335598568049e-05,
"loss": 0.0663,
"step": 216000
},
{
"epoch": 13.14,
"learning_rate": 1.3434166616103393e-05,
"loss": 0.0675,
"step": 216500
},
{
"epoch": 13.17,
"learning_rate": 1.3418997633638737e-05,
"loss": 0.067,
"step": 217000
},
{
"epoch": 13.2,
"learning_rate": 1.340382865117408e-05,
"loss": 0.0684,
"step": 217500
},
{
"epoch": 13.23,
"learning_rate": 1.3388690006674354e-05,
"loss": 0.067,
"step": 218000
},
{
"epoch": 13.26,
"learning_rate": 1.3373521024209697e-05,
"loss": 0.0678,
"step": 218500
},
{
"epoch": 13.29,
"learning_rate": 1.3358352041745042e-05,
"loss": 0.0658,
"step": 219000
},
{
"epoch": 13.32,
"learning_rate": 1.3343183059280385e-05,
"loss": 0.0657,
"step": 219500
},
{
"epoch": 13.35,
"learning_rate": 1.3328014076815727e-05,
"loss": 0.0687,
"step": 220000
},
{
"epoch": 13.38,
"learning_rate": 1.3312845094351072e-05,
"loss": 0.0667,
"step": 220500
},
{
"epoch": 13.41,
"learning_rate": 1.3297676111886416e-05,
"loss": 0.068,
"step": 221000
},
{
"epoch": 13.44,
"learning_rate": 1.328253746738669e-05,
"loss": 0.0692,
"step": 221500
},
{
"epoch": 13.47,
"learning_rate": 1.3267368484922032e-05,
"loss": 0.0688,
"step": 222000
},
{
"epoch": 13.5,
"learning_rate": 1.3252199502457377e-05,
"loss": 0.0675,
"step": 222500
},
{
"epoch": 13.53,
"learning_rate": 1.3237030519992719e-05,
"loss": 0.0688,
"step": 223000
},
{
"epoch": 13.56,
"learning_rate": 1.3221861537528065e-05,
"loss": 0.0676,
"step": 223500
},
{
"epoch": 13.59,
"learning_rate": 1.3206692555063408e-05,
"loss": 0.0681,
"step": 224000
},
{
"epoch": 13.62,
"learning_rate": 1.3191553910563682e-05,
"loss": 0.0676,
"step": 224500
},
{
"epoch": 13.65,
"learning_rate": 1.3176384928099024e-05,
"loss": 0.0709,
"step": 225000
},
{
"epoch": 13.68,
"learning_rate": 1.3161215945634367e-05,
"loss": 0.0693,
"step": 225500
},
{
"epoch": 13.71,
"learning_rate": 1.3146046963169713e-05,
"loss": 0.0686,
"step": 226000
},
{
"epoch": 13.74,
"learning_rate": 1.3130877980705055e-05,
"loss": 0.0693,
"step": 226500
},
{
"epoch": 13.77,
"learning_rate": 1.31157089982404e-05,
"loss": 0.0705,
"step": 227000
},
{
"epoch": 13.8,
"learning_rate": 1.3100540015775742e-05,
"loss": 0.0695,
"step": 227500
},
{
"epoch": 13.83,
"learning_rate": 1.3085401371276016e-05,
"loss": 0.0698,
"step": 228000
},
{
"epoch": 13.86,
"learning_rate": 1.3070232388811358e-05,
"loss": 0.0695,
"step": 228500
},
{
"epoch": 13.89,
"learning_rate": 1.3055063406346704e-05,
"loss": 0.0688,
"step": 229000
},
{
"epoch": 13.93,
"learning_rate": 1.3039894423882047e-05,
"loss": 0.0718,
"step": 229500
},
{
"epoch": 13.96,
"learning_rate": 1.3024725441417391e-05,
"loss": 0.0704,
"step": 230000
},
{
"epoch": 13.99,
"learning_rate": 1.3009556458952735e-05,
"loss": 0.0734,
"step": 230500
},
{
"epoch": 14.0,
"eval_bleu": 87.771,
"eval_gen_len": 14.7707,
"eval_loss": 0.16568207740783691,
"eval_runtime": 221.3957,
"eval_samples_per_second": 72.829,
"eval_steps_per_second": 2.276,
"step": 230734
},
{
"epoch": 14.02,
"learning_rate": 1.299441781445301e-05,
"loss": 0.066,
"step": 231000
},
{
"epoch": 14.05,
"learning_rate": 1.2979248831988352e-05,
"loss": 0.0608,
"step": 231500
},
{
"epoch": 14.08,
"learning_rate": 1.2964079849523694e-05,
"loss": 0.0603,
"step": 232000
},
{
"epoch": 14.11,
"learning_rate": 1.2948910867059039e-05,
"loss": 0.0606,
"step": 232500
},
{
"epoch": 14.14,
"learning_rate": 1.2933741884594381e-05,
"loss": 0.0612,
"step": 233000
},
{
"epoch": 14.17,
"learning_rate": 1.2918572902129727e-05,
"loss": 0.0618,
"step": 233500
},
{
"epoch": 14.2,
"learning_rate": 1.290340391966507e-05,
"loss": 0.0614,
"step": 234000
},
{
"epoch": 14.23,
"learning_rate": 1.2888234937200415e-05,
"loss": 0.0614,
"step": 234500
},
{
"epoch": 14.26,
"learning_rate": 1.2873096292700686e-05,
"loss": 0.061,
"step": 235000
},
{
"epoch": 14.29,
"learning_rate": 1.2857927310236032e-05,
"loss": 0.0638,
"step": 235500
},
{
"epoch": 14.32,
"learning_rate": 1.2842758327771374e-05,
"loss": 0.0615,
"step": 236000
},
{
"epoch": 14.35,
"learning_rate": 1.2827589345306717e-05,
"loss": 0.0637,
"step": 236500
},
{
"epoch": 14.38,
"learning_rate": 1.2812420362842061e-05,
"loss": 0.063,
"step": 237000
},
{
"epoch": 14.41,
"learning_rate": 1.2797281718342334e-05,
"loss": 0.0623,
"step": 237500
},
{
"epoch": 14.44,
"learning_rate": 1.2782112735877678e-05,
"loss": 0.0627,
"step": 238000
},
{
"epoch": 14.47,
"learning_rate": 1.2766943753413022e-05,
"loss": 0.0649,
"step": 238500
},
{
"epoch": 14.5,
"learning_rate": 1.2751774770948366e-05,
"loss": 0.0639,
"step": 239000
},
{
"epoch": 14.53,
"learning_rate": 1.2736605788483709e-05,
"loss": 0.0632,
"step": 239500
},
{
"epoch": 14.56,
"learning_rate": 1.2721467143983983e-05,
"loss": 0.064,
"step": 240000
},
{
"epoch": 14.59,
"learning_rate": 1.2706298161519325e-05,
"loss": 0.0641,
"step": 240500
},
{
"epoch": 14.62,
"learning_rate": 1.2691129179054671e-05,
"loss": 0.0639,
"step": 241000
},
{
"epoch": 14.65,
"learning_rate": 1.2675960196590014e-05,
"loss": 0.0645,
"step": 241500
},
{
"epoch": 14.68,
"learning_rate": 1.2660821552090288e-05,
"loss": 0.065,
"step": 242000
},
{
"epoch": 14.71,
"learning_rate": 1.264565256962563e-05,
"loss": 0.0653,
"step": 242500
},
{
"epoch": 14.74,
"learning_rate": 1.2630483587160973e-05,
"loss": 0.0646,
"step": 243000
},
{
"epoch": 14.77,
"learning_rate": 1.2615314604696319e-05,
"loss": 0.0658,
"step": 243500
},
{
"epoch": 14.8,
"learning_rate": 1.260017596019659e-05,
"loss": 0.0647,
"step": 244000
},
{
"epoch": 14.84,
"learning_rate": 1.2585006977731936e-05,
"loss": 0.0627,
"step": 244500
},
{
"epoch": 14.87,
"learning_rate": 1.2569837995267278e-05,
"loss": 0.0635,
"step": 245000
},
{
"epoch": 14.9,
"learning_rate": 1.2554669012802622e-05,
"loss": 0.0649,
"step": 245500
},
{
"epoch": 14.93,
"learning_rate": 1.2539500030337965e-05,
"loss": 0.0638,
"step": 246000
},
{
"epoch": 14.96,
"learning_rate": 1.252433104787331e-05,
"loss": 0.0652,
"step": 246500
},
{
"epoch": 14.99,
"learning_rate": 1.2509162065408653e-05,
"loss": 0.0649,
"step": 247000
},
{
"epoch": 15.0,
"eval_bleu": 87.6289,
"eval_gen_len": 14.7254,
"eval_loss": 0.1687079817056656,
"eval_runtime": 219.5287,
"eval_samples_per_second": 73.448,
"eval_steps_per_second": 2.296,
"step": 247215
},
{
"epoch": 15.02,
"learning_rate": 1.2494023420908927e-05,
"loss": 0.0556,
"step": 247500
},
{
"epoch": 15.05,
"learning_rate": 1.247885443844427e-05,
"loss": 0.0569,
"step": 248000
},
{
"epoch": 15.08,
"learning_rate": 1.2463685455979616e-05,
"loss": 0.0562,
"step": 248500
},
{
"epoch": 15.11,
"learning_rate": 1.2448516473514958e-05,
"loss": 0.0566,
"step": 249000
},
{
"epoch": 15.14,
"learning_rate": 1.24333474910503e-05,
"loss": 0.0575,
"step": 249500
},
{
"epoch": 15.17,
"learning_rate": 1.2418178508585645e-05,
"loss": 0.0576,
"step": 250000
},
{
"epoch": 15.2,
"learning_rate": 1.2403039864085917e-05,
"loss": 0.0577,
"step": 250500
},
{
"epoch": 15.23,
"learning_rate": 1.2387870881621262e-05,
"loss": 0.0562,
"step": 251000
},
{
"epoch": 15.26,
"learning_rate": 1.2372701899156606e-05,
"loss": 0.0566,
"step": 251500
},
{
"epoch": 15.29,
"learning_rate": 1.235753291669195e-05,
"loss": 0.0582,
"step": 252000
},
{
"epoch": 15.32,
"learning_rate": 1.2342363934227292e-05,
"loss": 0.0582,
"step": 252500
},
{
"epoch": 15.35,
"learning_rate": 1.2327194951762638e-05,
"loss": 0.0575,
"step": 253000
},
{
"epoch": 15.38,
"learning_rate": 1.231202596929798e-05,
"loss": 0.0593,
"step": 253500
},
{
"epoch": 15.41,
"learning_rate": 1.2296856986833323e-05,
"loss": 0.0584,
"step": 254000
},
{
"epoch": 15.44,
"learning_rate": 1.2281688004368667e-05,
"loss": 0.0583,
"step": 254500
},
{
"epoch": 15.47,
"learning_rate": 1.226654935986894e-05,
"loss": 0.058,
"step": 255000
},
{
"epoch": 15.5,
"learning_rate": 1.2251380377404284e-05,
"loss": 0.0594,
"step": 255500
},
{
"epoch": 15.53,
"learning_rate": 1.2236211394939628e-05,
"loss": 0.0601,
"step": 256000
},
{
"epoch": 15.56,
"learning_rate": 1.2221072750439903e-05,
"loss": 0.0582,
"step": 256500
},
{
"epoch": 15.59,
"learning_rate": 1.2205903767975245e-05,
"loss": 0.059,
"step": 257000
},
{
"epoch": 15.62,
"learning_rate": 1.219073478551059e-05,
"loss": 0.0574,
"step": 257500
},
{
"epoch": 15.65,
"learning_rate": 1.2175565803045932e-05,
"loss": 0.0589,
"step": 258000
},
{
"epoch": 15.68,
"learning_rate": 1.2160396820581278e-05,
"loss": 0.0586,
"step": 258500
},
{
"epoch": 15.72,
"learning_rate": 1.214522783811662e-05,
"loss": 0.0578,
"step": 259000
},
{
"epoch": 15.75,
"learning_rate": 1.2130058855651964e-05,
"loss": 0.0604,
"step": 259500
},
{
"epoch": 15.78,
"learning_rate": 1.2114889873187308e-05,
"loss": 0.059,
"step": 260000
},
{
"epoch": 15.81,
"learning_rate": 1.209975122868758e-05,
"loss": 0.0606,
"step": 260500
},
{
"epoch": 15.84,
"learning_rate": 1.2084582246222925e-05,
"loss": 0.0605,
"step": 261000
},
{
"epoch": 15.87,
"learning_rate": 1.2069443601723196e-05,
"loss": 0.0604,
"step": 261500
},
{
"epoch": 15.9,
"learning_rate": 1.2054274619258542e-05,
"loss": 0.0593,
"step": 262000
},
{
"epoch": 15.93,
"learning_rate": 1.2039105636793884e-05,
"loss": 0.0614,
"step": 262500
},
{
"epoch": 15.96,
"learning_rate": 1.2023936654329229e-05,
"loss": 0.0588,
"step": 263000
},
{
"epoch": 15.99,
"learning_rate": 1.2008767671864571e-05,
"loss": 0.0604,
"step": 263500
},
{
"epoch": 16.0,
"eval_bleu": 87.8903,
"eval_gen_len": 14.7974,
"eval_loss": 0.1746528446674347,
"eval_runtime": 222.3549,
"eval_samples_per_second": 72.515,
"eval_steps_per_second": 2.267,
"step": 263696
},
{
"epoch": 16.02,
"learning_rate": 1.1993598689399917e-05,
"loss": 0.0544,
"step": 264000
},
{
"epoch": 16.05,
"learning_rate": 1.197846004490019e-05,
"loss": 0.051,
"step": 264500
},
{
"epoch": 16.08,
"learning_rate": 1.1963291062435534e-05,
"loss": 0.0513,
"step": 265000
},
{
"epoch": 16.11,
"learning_rate": 1.1948122079970876e-05,
"loss": 0.0521,
"step": 265500
},
{
"epoch": 16.14,
"learning_rate": 1.1932953097506222e-05,
"loss": 0.0526,
"step": 266000
},
{
"epoch": 16.17,
"learning_rate": 1.1917784115041564e-05,
"loss": 0.0524,
"step": 266500
},
{
"epoch": 16.2,
"learning_rate": 1.1902615132576907e-05,
"loss": 0.0531,
"step": 267000
},
{
"epoch": 16.23,
"learning_rate": 1.1887446150112251e-05,
"loss": 0.0537,
"step": 267500
},
{
"epoch": 16.26,
"learning_rate": 1.1872277167647595e-05,
"loss": 0.0541,
"step": 268000
},
{
"epoch": 16.29,
"learning_rate": 1.1857168861112798e-05,
"loss": 0.0543,
"step": 268500
},
{
"epoch": 16.32,
"learning_rate": 1.184199987864814e-05,
"loss": 0.0537,
"step": 269000
},
{
"epoch": 16.35,
"learning_rate": 1.1826830896183486e-05,
"loss": 0.0534,
"step": 269500
},
{
"epoch": 16.38,
"learning_rate": 1.1811661913718829e-05,
"loss": 0.0527,
"step": 270000
},
{
"epoch": 16.41,
"learning_rate": 1.1796492931254173e-05,
"loss": 0.0535,
"step": 270500
},
{
"epoch": 16.44,
"learning_rate": 1.1781323948789515e-05,
"loss": 0.0545,
"step": 271000
},
{
"epoch": 16.47,
"learning_rate": 1.1766154966324861e-05,
"loss": 0.0537,
"step": 271500
},
{
"epoch": 16.5,
"learning_rate": 1.1750985983860204e-05,
"loss": 0.0538,
"step": 272000
},
{
"epoch": 16.53,
"learning_rate": 1.1735847339360478e-05,
"loss": 0.054,
"step": 272500
},
{
"epoch": 16.56,
"learning_rate": 1.1720708694860749e-05,
"loss": 0.055,
"step": 273000
},
{
"epoch": 16.59,
"learning_rate": 1.1705539712396093e-05,
"loss": 0.0544,
"step": 273500
},
{
"epoch": 16.63,
"learning_rate": 1.1690370729931437e-05,
"loss": 0.0557,
"step": 274000
},
{
"epoch": 16.66,
"learning_rate": 1.167520174746678e-05,
"loss": 0.0535,
"step": 274500
},
{
"epoch": 16.69,
"learning_rate": 1.1660032765002126e-05,
"loss": 0.0546,
"step": 275000
},
{
"epoch": 16.72,
"learning_rate": 1.1644894120502396e-05,
"loss": 0.0555,
"step": 275500
},
{
"epoch": 16.75,
"learning_rate": 1.1629725138037742e-05,
"loss": 0.0549,
"step": 276000
},
{
"epoch": 16.78,
"learning_rate": 1.1614556155573085e-05,
"loss": 0.0553,
"step": 276500
},
{
"epoch": 16.81,
"learning_rate": 1.1599387173108429e-05,
"loss": 0.0562,
"step": 277000
},
{
"epoch": 16.84,
"learning_rate": 1.1584218190643773e-05,
"loss": 0.0557,
"step": 277500
},
{
"epoch": 16.87,
"learning_rate": 1.1569049208179117e-05,
"loss": 0.0555,
"step": 278000
},
{
"epoch": 16.9,
"learning_rate": 1.155388022571446e-05,
"loss": 0.0566,
"step": 278500
},
{
"epoch": 16.93,
"learning_rate": 1.1538711243249802e-05,
"loss": 0.0571,
"step": 279000
},
{
"epoch": 16.96,
"learning_rate": 1.1523572598750076e-05,
"loss": 0.0558,
"step": 279500
},
{
"epoch": 16.99,
"learning_rate": 1.1508403616285419e-05,
"loss": 0.0555,
"step": 280000
},
{
"epoch": 17.0,
"eval_bleu": 87.7539,
"eval_gen_len": 14.7936,
"eval_loss": 0.17870613932609558,
"eval_runtime": 227.9238,
"eval_samples_per_second": 70.743,
"eval_steps_per_second": 2.211,
"step": 280177
},
{
"epoch": 17.02,
"learning_rate": 1.1493234633820765e-05,
"loss": 0.0499,
"step": 280500
},
{
"epoch": 17.05,
"learning_rate": 1.1478065651356107e-05,
"loss": 0.0476,
"step": 281000
},
{
"epoch": 17.08,
"learning_rate": 1.1462896668891452e-05,
"loss": 0.0495,
"step": 281500
},
{
"epoch": 17.11,
"learning_rate": 1.1447727686426796e-05,
"loss": 0.0482,
"step": 282000
},
{
"epoch": 17.14,
"learning_rate": 1.143258904192707e-05,
"loss": 0.0475,
"step": 282500
},
{
"epoch": 17.17,
"learning_rate": 1.1417420059462412e-05,
"loss": 0.0499,
"step": 283000
},
{
"epoch": 17.2,
"learning_rate": 1.1402251076997757e-05,
"loss": 0.049,
"step": 283500
},
{
"epoch": 17.23,
"learning_rate": 1.1387082094533099e-05,
"loss": 0.0481,
"step": 284000
},
{
"epoch": 17.26,
"learning_rate": 1.1371913112068442e-05,
"loss": 0.0479,
"step": 284500
},
{
"epoch": 17.29,
"learning_rate": 1.1356774467568716e-05,
"loss": 0.0486,
"step": 285000
},
{
"epoch": 17.32,
"learning_rate": 1.134163582306899e-05,
"loss": 0.0501,
"step": 285500
},
{
"epoch": 17.35,
"learning_rate": 1.1326466840604332e-05,
"loss": 0.051,
"step": 286000
},
{
"epoch": 17.38,
"learning_rate": 1.1311297858139677e-05,
"loss": 0.0487,
"step": 286500
},
{
"epoch": 17.41,
"learning_rate": 1.1296128875675021e-05,
"loss": 0.0504,
"step": 287000
},
{
"epoch": 17.44,
"learning_rate": 1.1280959893210363e-05,
"loss": 0.0503,
"step": 287500
},
{
"epoch": 17.47,
"learning_rate": 1.126579091074571e-05,
"loss": 0.0499,
"step": 288000
},
{
"epoch": 17.51,
"learning_rate": 1.1250621928281052e-05,
"loss": 0.0517,
"step": 288500
},
{
"epoch": 17.54,
"learning_rate": 1.1235452945816396e-05,
"loss": 0.0504,
"step": 289000
},
{
"epoch": 17.57,
"learning_rate": 1.1220283963351738e-05,
"loss": 0.0499,
"step": 289500
},
{
"epoch": 17.6,
"learning_rate": 1.1205114980887084e-05,
"loss": 0.0525,
"step": 290000
},
{
"epoch": 17.63,
"learning_rate": 1.1189945998422427e-05,
"loss": 0.0496,
"step": 290500
},
{
"epoch": 17.66,
"learning_rate": 1.1174807353922701e-05,
"loss": 0.0498,
"step": 291000
},
{
"epoch": 17.69,
"learning_rate": 1.1159638371458043e-05,
"loss": 0.0513,
"step": 291500
},
{
"epoch": 17.72,
"learning_rate": 1.1144469388993386e-05,
"loss": 0.0501,
"step": 292000
},
{
"epoch": 17.75,
"learning_rate": 1.1129300406528732e-05,
"loss": 0.0511,
"step": 292500
},
{
"epoch": 17.78,
"learning_rate": 1.1114131424064074e-05,
"loss": 0.0503,
"step": 293000
},
{
"epoch": 17.81,
"learning_rate": 1.1098962441599419e-05,
"loss": 0.0515,
"step": 293500
},
{
"epoch": 17.84,
"learning_rate": 1.1083793459134761e-05,
"loss": 0.0527,
"step": 294000
},
{
"epoch": 17.87,
"learning_rate": 1.1068624476670107e-05,
"loss": 0.0493,
"step": 294500
},
{
"epoch": 17.9,
"learning_rate": 1.105348583217038e-05,
"loss": 0.0503,
"step": 295000
},
{
"epoch": 17.93,
"learning_rate": 1.1038316849705724e-05,
"loss": 0.0506,
"step": 295500
},
{
"epoch": 17.96,
"learning_rate": 1.1023208543170924e-05,
"loss": 0.0508,
"step": 296000
},
{
"epoch": 17.99,
"learning_rate": 1.100803956070627e-05,
"loss": 0.051,
"step": 296500
},
{
"epoch": 18.0,
"eval_bleu": 87.7359,
"eval_gen_len": 14.7867,
"eval_loss": 0.18140269815921783,
"eval_runtime": 217.161,
"eval_samples_per_second": 74.249,
"eval_steps_per_second": 2.321,
"step": 296658
},
{
"epoch": 18.02,
"learning_rate": 1.0992870578241613e-05,
"loss": 0.0467,
"step": 297000
},
{
"epoch": 18.05,
"learning_rate": 1.0977701595776957e-05,
"loss": 0.0441,
"step": 297500
},
{
"epoch": 18.08,
"learning_rate": 1.09625326133123e-05,
"loss": 0.0444,
"step": 298000
},
{
"epoch": 18.11,
"learning_rate": 1.0947363630847642e-05,
"loss": 0.0444,
"step": 298500
},
{
"epoch": 18.14,
"learning_rate": 1.0932194648382988e-05,
"loss": 0.0448,
"step": 299000
},
{
"epoch": 18.17,
"learning_rate": 1.091702566591833e-05,
"loss": 0.0446,
"step": 299500
},
{
"epoch": 18.2,
"learning_rate": 1.0901856683453676e-05,
"loss": 0.0451,
"step": 300000
},
{
"epoch": 18.23,
"learning_rate": 1.0886687700989019e-05,
"loss": 0.0452,
"step": 300500
},
{
"epoch": 18.26,
"learning_rate": 1.0871518718524363e-05,
"loss": 0.0457,
"step": 301000
},
{
"epoch": 18.29,
"learning_rate": 1.0856349736059705e-05,
"loss": 0.0463,
"step": 301500
},
{
"epoch": 18.32,
"learning_rate": 1.0841180753595048e-05,
"loss": 0.0451,
"step": 302000
},
{
"epoch": 18.35,
"learning_rate": 1.082610278502518e-05,
"loss": 0.0457,
"step": 302500
},
{
"epoch": 18.38,
"learning_rate": 1.0810933802560526e-05,
"loss": 0.0462,
"step": 303000
},
{
"epoch": 18.42,
"learning_rate": 1.0795764820095869e-05,
"loss": 0.0456,
"step": 303500
},
{
"epoch": 18.45,
"learning_rate": 1.0780595837631213e-05,
"loss": 0.0459,
"step": 304000
},
{
"epoch": 18.48,
"learning_rate": 1.0765457193131485e-05,
"loss": 0.0475,
"step": 304500
},
{
"epoch": 18.51,
"learning_rate": 1.0750288210666828e-05,
"loss": 0.047,
"step": 305000
},
{
"epoch": 18.54,
"learning_rate": 1.0735119228202174e-05,
"loss": 0.0458,
"step": 305500
},
{
"epoch": 18.57,
"learning_rate": 1.0719950245737516e-05,
"loss": 0.0471,
"step": 306000
},
{
"epoch": 18.6,
"learning_rate": 1.070478126327286e-05,
"loss": 0.0464,
"step": 306500
},
{
"epoch": 18.63,
"learning_rate": 1.0689612280808203e-05,
"loss": 0.0472,
"step": 307000
},
{
"epoch": 18.66,
"learning_rate": 1.0674443298343549e-05,
"loss": 0.0468,
"step": 307500
},
{
"epoch": 18.69,
"learning_rate": 1.0659274315878891e-05,
"loss": 0.0463,
"step": 308000
},
{
"epoch": 18.72,
"learning_rate": 1.0644135671379166e-05,
"loss": 0.0473,
"step": 308500
},
{
"epoch": 18.75,
"learning_rate": 1.0628966688914508e-05,
"loss": 0.0472,
"step": 309000
},
{
"epoch": 18.78,
"learning_rate": 1.0613797706449854e-05,
"loss": 0.0456,
"step": 309500
},
{
"epoch": 18.81,
"learning_rate": 1.0598628723985196e-05,
"loss": 0.0474,
"step": 310000
},
{
"epoch": 18.84,
"learning_rate": 1.0583459741520539e-05,
"loss": 0.0466,
"step": 310500
},
{
"epoch": 18.87,
"learning_rate": 1.0568290759055883e-05,
"loss": 0.0471,
"step": 311000
},
{
"epoch": 18.9,
"learning_rate": 1.0553121776591226e-05,
"loss": 0.0477,
"step": 311500
},
{
"epoch": 18.93,
"learning_rate": 1.0537952794126572e-05,
"loss": 0.0472,
"step": 312000
},
{
"epoch": 18.96,
"learning_rate": 1.0522783811661914e-05,
"loss": 0.0469,
"step": 312500
},
{
"epoch": 18.99,
"learning_rate": 1.0507645167162188e-05,
"loss": 0.0466,
"step": 313000
},
{
"epoch": 19.0,
"eval_bleu": 87.7646,
"eval_gen_len": 14.7869,
"eval_loss": 0.18415296077728271,
"eval_runtime": 217.4372,
"eval_samples_per_second": 74.155,
"eval_steps_per_second": 2.318,
"step": 313139
},
{
"epoch": 19.02,
"learning_rate": 1.049247618469753e-05,
"loss": 0.0435,
"step": 313500
},
{
"epoch": 19.05,
"learning_rate": 1.0477307202232877e-05,
"loss": 0.0413,
"step": 314000
},
{
"epoch": 19.08,
"learning_rate": 1.0462168557733147e-05,
"loss": 0.0409,
"step": 314500
},
{
"epoch": 19.11,
"learning_rate": 1.0446999575268493e-05,
"loss": 0.0408,
"step": 315000
},
{
"epoch": 19.14,
"learning_rate": 1.0431830592803836e-05,
"loss": 0.0414,
"step": 315500
},
{
"epoch": 19.17,
"learning_rate": 1.0416661610339178e-05,
"loss": 0.0416,
"step": 316000
},
{
"epoch": 19.2,
"learning_rate": 1.0401492627874522e-05,
"loss": 0.0417,
"step": 316500
},
{
"epoch": 19.23,
"learning_rate": 1.0386323645409867e-05,
"loss": 0.0422,
"step": 317000
},
{
"epoch": 19.26,
"learning_rate": 1.037115466294521e-05,
"loss": 0.0424,
"step": 317500
},
{
"epoch": 19.29,
"learning_rate": 1.0355985680480553e-05,
"loss": 0.0412,
"step": 318000
},
{
"epoch": 19.33,
"learning_rate": 1.03408166980159e-05,
"loss": 0.0422,
"step": 318500
},
{
"epoch": 19.36,
"learning_rate": 1.0325647715551242e-05,
"loss": 0.0401,
"step": 319000
},
{
"epoch": 19.39,
"learning_rate": 1.0310478733086586e-05,
"loss": 0.0411,
"step": 319500
},
{
"epoch": 19.42,
"learning_rate": 1.0295340088586858e-05,
"loss": 0.0428,
"step": 320000
},
{
"epoch": 19.45,
"learning_rate": 1.0280171106122203e-05,
"loss": 0.0443,
"step": 320500
},
{
"epoch": 19.48,
"learning_rate": 1.0265002123657545e-05,
"loss": 0.0435,
"step": 321000
},
{
"epoch": 19.51,
"learning_rate": 1.024983314119289e-05,
"loss": 0.0412,
"step": 321500
},
{
"epoch": 19.54,
"learning_rate": 1.0234664158728233e-05,
"loss": 0.0433,
"step": 322000
},
{
"epoch": 19.57,
"learning_rate": 1.0219525514228506e-05,
"loss": 0.0425,
"step": 322500
},
{
"epoch": 19.6,
"learning_rate": 1.020435653176385e-05,
"loss": 0.0428,
"step": 323000
},
{
"epoch": 19.63,
"learning_rate": 1.0189187549299193e-05,
"loss": 0.0432,
"step": 323500
},
{
"epoch": 19.66,
"learning_rate": 1.0174018566834538e-05,
"loss": 0.0428,
"step": 324000
},
{
"epoch": 19.69,
"learning_rate": 1.0158849584369881e-05,
"loss": 0.0435,
"step": 324500
},
{
"epoch": 19.72,
"learning_rate": 1.0143680601905225e-05,
"loss": 0.0433,
"step": 325000
},
{
"epoch": 19.75,
"learning_rate": 1.012851161944057e-05,
"loss": 0.0433,
"step": 325500
},
{
"epoch": 19.78,
"learning_rate": 1.0113342636975914e-05,
"loss": 0.0441,
"step": 326000
},
{
"epoch": 19.81,
"learning_rate": 1.0098203992476186e-05,
"loss": 0.0434,
"step": 326500
},
{
"epoch": 19.84,
"learning_rate": 1.0083035010011529e-05,
"loss": 0.0425,
"step": 327000
},
{
"epoch": 19.87,
"learning_rate": 1.0067866027546873e-05,
"loss": 0.0442,
"step": 327500
},
{
"epoch": 19.9,
"learning_rate": 1.0052697045082215e-05,
"loss": 0.0438,
"step": 328000
},
{
"epoch": 19.93,
"learning_rate": 1.0037528062617561e-05,
"loss": 0.0435,
"step": 328500
},
{
"epoch": 19.96,
"learning_rate": 1.0022359080152904e-05,
"loss": 0.0436,
"step": 329000
},
{
"epoch": 19.99,
"learning_rate": 1.0007220435653178e-05,
"loss": 0.046,
"step": 329500
},
{
"epoch": 20.0,
"eval_bleu": 87.8274,
"eval_gen_len": 14.7548,
"eval_loss": 0.1875508427619934,
"eval_runtime": 216.5355,
"eval_samples_per_second": 74.464,
"eval_steps_per_second": 2.328,
"step": 329620
},
{
"epoch": 20.02,
"learning_rate": 9.99205145318852e-06,
"loss": 0.0389,
"step": 330000
},
{
"epoch": 20.05,
"learning_rate": 9.976882470723864e-06,
"loss": 0.0373,
"step": 330500
},
{
"epoch": 20.08,
"learning_rate": 9.961713488259209e-06,
"loss": 0.0381,
"step": 331000
},
{
"epoch": 20.11,
"learning_rate": 9.946544505794551e-06,
"loss": 0.0383,
"step": 331500
},
{
"epoch": 20.14,
"learning_rate": 9.931375523329895e-06,
"loss": 0.039,
"step": 332000
},
{
"epoch": 20.17,
"learning_rate": 9.91623687883017e-06,
"loss": 0.039,
"step": 332500
},
{
"epoch": 20.21,
"learning_rate": 9.901067896365512e-06,
"loss": 0.0385,
"step": 333000
},
{
"epoch": 20.24,
"learning_rate": 9.885898913900856e-06,
"loss": 0.0387,
"step": 333500
},
{
"epoch": 20.27,
"learning_rate": 9.8707299314362e-06,
"loss": 0.0384,
"step": 334000
},
{
"epoch": 20.3,
"learning_rate": 9.855560948971545e-06,
"loss": 0.0393,
"step": 334500
},
{
"epoch": 20.33,
"learning_rate": 9.840391966506887e-06,
"loss": 0.0393,
"step": 335000
},
{
"epoch": 20.36,
"learning_rate": 9.825222984042231e-06,
"loss": 0.038,
"step": 335500
},
{
"epoch": 20.39,
"learning_rate": 9.810084339542504e-06,
"loss": 0.04,
"step": 336000
},
{
"epoch": 20.42,
"learning_rate": 9.794915357077848e-06,
"loss": 0.04,
"step": 336500
},
{
"epoch": 20.45,
"learning_rate": 9.779746374613192e-06,
"loss": 0.0397,
"step": 337000
},
{
"epoch": 20.48,
"learning_rate": 9.764577392148535e-06,
"loss": 0.0396,
"step": 337500
},
{
"epoch": 20.51,
"learning_rate": 9.749408409683879e-06,
"loss": 0.0386,
"step": 338000
},
{
"epoch": 20.54,
"learning_rate": 9.734239427219223e-06,
"loss": 0.0391,
"step": 338500
},
{
"epoch": 20.57,
"learning_rate": 9.719070444754567e-06,
"loss": 0.0404,
"step": 339000
},
{
"epoch": 20.6,
"learning_rate": 9.703901462289911e-06,
"loss": 0.04,
"step": 339500
},
{
"epoch": 20.63,
"learning_rate": 9.688762817790184e-06,
"loss": 0.0407,
"step": 340000
},
{
"epoch": 20.66,
"learning_rate": 9.673593835325528e-06,
"loss": 0.04,
"step": 340500
},
{
"epoch": 20.69,
"learning_rate": 9.65842485286087e-06,
"loss": 0.0399,
"step": 341000
},
{
"epoch": 20.72,
"learning_rate": 9.643255870396215e-06,
"loss": 0.0381,
"step": 341500
},
{
"epoch": 20.75,
"learning_rate": 9.628117225896487e-06,
"loss": 0.0395,
"step": 342000
},
{
"epoch": 20.78,
"learning_rate": 9.612948243431831e-06,
"loss": 0.0397,
"step": 342500
},
{
"epoch": 20.81,
"learning_rate": 9.597809598932104e-06,
"loss": 0.0407,
"step": 343000
},
{
"epoch": 20.84,
"learning_rate": 9.582640616467448e-06,
"loss": 0.04,
"step": 343500
},
{
"epoch": 20.87,
"learning_rate": 9.567471634002792e-06,
"loss": 0.0411,
"step": 344000
},
{
"epoch": 20.9,
"learning_rate": 9.552302651538135e-06,
"loss": 0.0412,
"step": 344500
},
{
"epoch": 20.93,
"learning_rate": 9.537133669073479e-06,
"loss": 0.0396,
"step": 345000
},
{
"epoch": 20.96,
"learning_rate": 9.521964686608823e-06,
"loss": 0.0407,
"step": 345500
},
{
"epoch": 20.99,
"learning_rate": 9.506826042109096e-06,
"loss": 0.0402,
"step": 346000
},
{
"epoch": 21.0,
"eval_bleu": 87.6913,
"eval_gen_len": 14.761,
"eval_loss": 0.19335225224494934,
"eval_runtime": 217.2279,
"eval_samples_per_second": 74.226,
"eval_steps_per_second": 2.32,
"step": 346101
},
{
"epoch": 21.02,
"learning_rate": 9.49168739760937e-06,
"loss": 0.0358,
"step": 346500
},
{
"epoch": 21.05,
"learning_rate": 9.476518415144712e-06,
"loss": 0.034,
"step": 347000
},
{
"epoch": 21.08,
"learning_rate": 9.461349432680057e-06,
"loss": 0.0357,
"step": 347500
},
{
"epoch": 21.12,
"learning_rate": 9.4461804502154e-06,
"loss": 0.0359,
"step": 348000
},
{
"epoch": 21.15,
"learning_rate": 9.431011467750743e-06,
"loss": 0.0361,
"step": 348500
},
{
"epoch": 21.18,
"learning_rate": 9.415842485286087e-06,
"loss": 0.0361,
"step": 349000
},
{
"epoch": 21.21,
"learning_rate": 9.400673502821432e-06,
"loss": 0.0359,
"step": 349500
},
{
"epoch": 21.24,
"learning_rate": 9.385504520356776e-06,
"loss": 0.0349,
"step": 350000
},
{
"epoch": 21.27,
"learning_rate": 9.370335537892118e-06,
"loss": 0.0363,
"step": 350500
},
{
"epoch": 21.3,
"learning_rate": 9.355166555427463e-06,
"loss": 0.0351,
"step": 351000
},
{
"epoch": 21.33,
"learning_rate": 9.340027910927735e-06,
"loss": 0.0365,
"step": 351500
},
{
"epoch": 21.36,
"learning_rate": 9.32488926642801e-06,
"loss": 0.0361,
"step": 352000
},
{
"epoch": 21.39,
"learning_rate": 9.309720283963353e-06,
"loss": 0.0367,
"step": 352500
},
{
"epoch": 21.42,
"learning_rate": 9.294551301498696e-06,
"loss": 0.0355,
"step": 353000
},
{
"epoch": 21.45,
"learning_rate": 9.27938231903404e-06,
"loss": 0.0364,
"step": 353500
},
{
"epoch": 21.48,
"learning_rate": 9.264213336569383e-06,
"loss": 0.0361,
"step": 354000
},
{
"epoch": 21.51,
"learning_rate": 9.249044354104727e-06,
"loss": 0.0367,
"step": 354500
},
{
"epoch": 21.54,
"learning_rate": 9.233905709605e-06,
"loss": 0.0372,
"step": 355000
},
{
"epoch": 21.57,
"learning_rate": 9.218736727140343e-06,
"loss": 0.0371,
"step": 355500
},
{
"epoch": 21.6,
"learning_rate": 9.203567744675688e-06,
"loss": 0.037,
"step": 356000
},
{
"epoch": 21.63,
"learning_rate": 9.188398762211032e-06,
"loss": 0.0376,
"step": 356500
},
{
"epoch": 21.66,
"learning_rate": 9.173229779746376e-06,
"loss": 0.0369,
"step": 357000
},
{
"epoch": 21.69,
"learning_rate": 9.158060797281719e-06,
"loss": 0.0379,
"step": 357500
},
{
"epoch": 21.72,
"learning_rate": 9.142891814817063e-06,
"loss": 0.0372,
"step": 358000
},
{
"epoch": 21.75,
"learning_rate": 9.127722832352407e-06,
"loss": 0.0375,
"step": 358500
},
{
"epoch": 21.78,
"learning_rate": 9.112553849887751e-06,
"loss": 0.038,
"step": 359000
},
{
"epoch": 21.81,
"learning_rate": 9.097384867423094e-06,
"loss": 0.0376,
"step": 359500
},
{
"epoch": 21.84,
"learning_rate": 9.082215884958438e-06,
"loss": 0.0366,
"step": 360000
},
{
"epoch": 21.87,
"learning_rate": 9.067046902493782e-06,
"loss": 0.038,
"step": 360500
},
{
"epoch": 21.9,
"learning_rate": 9.051908257994054e-06,
"loss": 0.0375,
"step": 361000
},
{
"epoch": 21.93,
"learning_rate": 9.036739275529399e-06,
"loss": 0.0363,
"step": 361500
},
{
"epoch": 21.96,
"learning_rate": 9.021570293064741e-06,
"loss": 0.0369,
"step": 362000
},
{
"epoch": 22.0,
"learning_rate": 9.006401310600085e-06,
"loss": 0.0374,
"step": 362500
},
{
"epoch": 22.0,
"eval_bleu": 87.7892,
"eval_gen_len": 14.7636,
"eval_loss": 0.19496643543243408,
"eval_runtime": 216.9414,
"eval_samples_per_second": 74.324,
"eval_steps_per_second": 2.323,
"step": 362582
},
{
"epoch": 22.03,
"learning_rate": 8.99123232813543e-06,
"loss": 0.0331,
"step": 363000
},
{
"epoch": 22.06,
"learning_rate": 8.976093683635702e-06,
"loss": 0.0317,
"step": 363500
},
{
"epoch": 22.09,
"learning_rate": 8.960924701171046e-06,
"loss": 0.0319,
"step": 364000
},
{
"epoch": 22.12,
"learning_rate": 8.94575571870639e-06,
"loss": 0.0338,
"step": 364500
},
{
"epoch": 22.15,
"learning_rate": 8.930586736241733e-06,
"loss": 0.0331,
"step": 365000
},
{
"epoch": 22.18,
"learning_rate": 8.915417753777077e-06,
"loss": 0.0339,
"step": 365500
},
{
"epoch": 22.21,
"learning_rate": 8.90030944724228e-06,
"loss": 0.0336,
"step": 366000
},
{
"epoch": 22.24,
"learning_rate": 8.885140464777622e-06,
"loss": 0.0325,
"step": 366500
},
{
"epoch": 22.27,
"learning_rate": 8.869971482312966e-06,
"loss": 0.0324,
"step": 367000
},
{
"epoch": 22.3,
"learning_rate": 8.85480249984831e-06,
"loss": 0.0336,
"step": 367500
},
{
"epoch": 22.33,
"learning_rate": 8.839633517383655e-06,
"loss": 0.0336,
"step": 368000
},
{
"epoch": 22.36,
"learning_rate": 8.824464534918999e-06,
"loss": 0.0333,
"step": 368500
},
{
"epoch": 22.39,
"learning_rate": 8.809295552454343e-06,
"loss": 0.0335,
"step": 369000
},
{
"epoch": 22.42,
"learning_rate": 8.794126569989686e-06,
"loss": 0.0327,
"step": 369500
},
{
"epoch": 22.45,
"learning_rate": 8.77895758752503e-06,
"loss": 0.0348,
"step": 370000
},
{
"epoch": 22.48,
"learning_rate": 8.763788605060374e-06,
"loss": 0.0348,
"step": 370500
},
{
"epoch": 22.51,
"learning_rate": 8.748649960560646e-06,
"loss": 0.0343,
"step": 371000
},
{
"epoch": 22.54,
"learning_rate": 8.733480978095989e-06,
"loss": 0.0345,
"step": 371500
},
{
"epoch": 22.57,
"learning_rate": 8.718311995631333e-06,
"loss": 0.0334,
"step": 372000
},
{
"epoch": 22.6,
"learning_rate": 8.703143013166677e-06,
"loss": 0.0341,
"step": 372500
},
{
"epoch": 22.63,
"learning_rate": 8.687974030702021e-06,
"loss": 0.0342,
"step": 373000
},
{
"epoch": 22.66,
"learning_rate": 8.672805048237366e-06,
"loss": 0.0336,
"step": 373500
},
{
"epoch": 22.69,
"learning_rate": 8.657636065772708e-06,
"loss": 0.0343,
"step": 374000
},
{
"epoch": 22.72,
"learning_rate": 8.642467083308052e-06,
"loss": 0.0341,
"step": 374500
},
{
"epoch": 22.75,
"learning_rate": 8.627298100843396e-06,
"loss": 0.0339,
"step": 375000
},
{
"epoch": 22.78,
"learning_rate": 8.612159456343669e-06,
"loss": 0.0345,
"step": 375500
},
{
"epoch": 22.81,
"learning_rate": 8.596990473879013e-06,
"loss": 0.036,
"step": 376000
},
{
"epoch": 22.84,
"learning_rate": 8.581821491414357e-06,
"loss": 0.0348,
"step": 376500
},
{
"epoch": 22.87,
"learning_rate": 8.5666525089497e-06,
"loss": 0.0348,
"step": 377000
},
{
"epoch": 22.91,
"learning_rate": 8.551513864449972e-06,
"loss": 0.0343,
"step": 377500
},
{
"epoch": 22.94,
"learning_rate": 8.536344881985317e-06,
"loss": 0.036,
"step": 378000
},
{
"epoch": 22.97,
"learning_rate": 8.52117589952066e-06,
"loss": 0.0356,
"step": 378500
},
{
"epoch": 23.0,
"learning_rate": 8.506006917056005e-06,
"loss": 0.035,
"step": 379000
},
{
"epoch": 23.0,
"eval_bleu": 87.8202,
"eval_gen_len": 14.7439,
"eval_loss": 0.1992015838623047,
"eval_runtime": 216.913,
"eval_samples_per_second": 74.334,
"eval_steps_per_second": 2.324,
"step": 379063
},
{
"epoch": 23.03,
"learning_rate": 8.490837934591349e-06,
"loss": 0.0311,
"step": 379500
},
{
"epoch": 23.06,
"learning_rate": 8.475699290091622e-06,
"loss": 0.0301,
"step": 380000
},
{
"epoch": 23.09,
"learning_rate": 8.460530307626966e-06,
"loss": 0.0302,
"step": 380500
},
{
"epoch": 23.12,
"learning_rate": 8.445361325162308e-06,
"loss": 0.0295,
"step": 381000
},
{
"epoch": 23.15,
"learning_rate": 8.430192342697652e-06,
"loss": 0.0303,
"step": 381500
},
{
"epoch": 23.18,
"learning_rate": 8.415023360232997e-06,
"loss": 0.0308,
"step": 382000
},
{
"epoch": 23.21,
"learning_rate": 8.39988471573327e-06,
"loss": 0.0304,
"step": 382500
},
{
"epoch": 23.24,
"learning_rate": 8.384715733268613e-06,
"loss": 0.0309,
"step": 383000
},
{
"epoch": 23.27,
"learning_rate": 8.369577088768886e-06,
"loss": 0.031,
"step": 383500
},
{
"epoch": 23.3,
"learning_rate": 8.35443844426916e-06,
"loss": 0.0306,
"step": 384000
},
{
"epoch": 23.33,
"learning_rate": 8.339269461804503e-06,
"loss": 0.0315,
"step": 384500
},
{
"epoch": 23.36,
"learning_rate": 8.324100479339847e-06,
"loss": 0.0323,
"step": 385000
},
{
"epoch": 23.39,
"learning_rate": 8.30893149687519e-06,
"loss": 0.0324,
"step": 385500
},
{
"epoch": 23.42,
"learning_rate": 8.293762514410533e-06,
"loss": 0.032,
"step": 386000
},
{
"epoch": 23.45,
"learning_rate": 8.278593531945878e-06,
"loss": 0.0312,
"step": 386500
},
{
"epoch": 23.48,
"learning_rate": 8.263424549481222e-06,
"loss": 0.0318,
"step": 387000
},
{
"epoch": 23.51,
"learning_rate": 8.248255567016566e-06,
"loss": 0.032,
"step": 387500
},
{
"epoch": 23.54,
"learning_rate": 8.233086584551908e-06,
"loss": 0.0317,
"step": 388000
},
{
"epoch": 23.57,
"learning_rate": 8.217917602087253e-06,
"loss": 0.0319,
"step": 388500
},
{
"epoch": 23.6,
"learning_rate": 8.202748619622597e-06,
"loss": 0.0317,
"step": 389000
},
{
"epoch": 23.63,
"learning_rate": 8.18760997512287e-06,
"loss": 0.0311,
"step": 389500
},
{
"epoch": 23.66,
"learning_rate": 8.172440992658212e-06,
"loss": 0.0323,
"step": 390000
},
{
"epoch": 23.69,
"learning_rate": 8.157272010193556e-06,
"loss": 0.0316,
"step": 390500
},
{
"epoch": 23.72,
"learning_rate": 8.1421030277289e-06,
"loss": 0.0304,
"step": 391000
},
{
"epoch": 23.75,
"learning_rate": 8.126964383229173e-06,
"loss": 0.0315,
"step": 391500
},
{
"epoch": 23.78,
"learning_rate": 8.111795400764517e-06,
"loss": 0.0322,
"step": 392000
},
{
"epoch": 23.82,
"learning_rate": 8.096626418299861e-06,
"loss": 0.0328,
"step": 392500
},
{
"epoch": 23.85,
"learning_rate": 8.081457435835205e-06,
"loss": 0.032,
"step": 393000
},
{
"epoch": 23.88,
"learning_rate": 8.06628845337055e-06,
"loss": 0.0315,
"step": 393500
},
{
"epoch": 23.91,
"learning_rate": 8.051119470905892e-06,
"loss": 0.0326,
"step": 394000
},
{
"epoch": 23.94,
"learning_rate": 8.035950488441236e-06,
"loss": 0.0317,
"step": 394500
},
{
"epoch": 23.97,
"learning_rate": 8.020781505976579e-06,
"loss": 0.0318,
"step": 395000
},
{
"epoch": 24.0,
"learning_rate": 8.005612523511923e-06,
"loss": 0.0333,
"step": 395500
},
{
"epoch": 24.0,
"eval_bleu": 87.6814,
"eval_gen_len": 14.7444,
"eval_loss": 0.20281654596328735,
"eval_runtime": 216.1907,
"eval_samples_per_second": 74.582,
"eval_steps_per_second": 2.331,
"step": 395544
},
{
"epoch": 24.03,
"learning_rate": 7.990473879012195e-06,
"loss": 0.0286,
"step": 396000
},
{
"epoch": 24.06,
"learning_rate": 7.97530489654754e-06,
"loss": 0.0284,
"step": 396500
},
{
"epoch": 24.09,
"learning_rate": 7.960135914082884e-06,
"loss": 0.0281,
"step": 397000
},
{
"epoch": 24.12,
"learning_rate": 7.944966931618228e-06,
"loss": 0.0278,
"step": 397500
},
{
"epoch": 24.15,
"learning_rate": 7.9298282871185e-06,
"loss": 0.0283,
"step": 398000
},
{
"epoch": 24.18,
"learning_rate": 7.914659304653845e-06,
"loss": 0.0279,
"step": 398500
},
{
"epoch": 24.21,
"learning_rate": 7.899490322189189e-06,
"loss": 0.0286,
"step": 399000
},
{
"epoch": 24.24,
"learning_rate": 7.884321339724533e-06,
"loss": 0.0283,
"step": 399500
},
{
"epoch": 24.27,
"learning_rate": 7.869182695224805e-06,
"loss": 0.0296,
"step": 400000
},
{
"epoch": 24.3,
"learning_rate": 7.85401371276015e-06,
"loss": 0.0285,
"step": 400500
},
{
"epoch": 24.33,
"learning_rate": 7.838844730295492e-06,
"loss": 0.0292,
"step": 401000
},
{
"epoch": 24.36,
"learning_rate": 7.823675747830836e-06,
"loss": 0.0288,
"step": 401500
},
{
"epoch": 24.39,
"learning_rate": 7.808506765366179e-06,
"loss": 0.0286,
"step": 402000
},
{
"epoch": 24.42,
"learning_rate": 7.793337782901523e-06,
"loss": 0.03,
"step": 402500
},
{
"epoch": 24.45,
"learning_rate": 7.778168800436867e-06,
"loss": 0.03,
"step": 403000
},
{
"epoch": 24.48,
"learning_rate": 7.762999817972211e-06,
"loss": 0.0294,
"step": 403500
},
{
"epoch": 24.51,
"learning_rate": 7.747830835507556e-06,
"loss": 0.0288,
"step": 404000
},
{
"epoch": 24.54,
"learning_rate": 7.732661853042898e-06,
"loss": 0.0287,
"step": 404500
},
{
"epoch": 24.57,
"learning_rate": 7.717492870578242e-06,
"loss": 0.0304,
"step": 405000
},
{
"epoch": 24.6,
"learning_rate": 7.702323888113586e-06,
"loss": 0.0306,
"step": 405500
},
{
"epoch": 24.63,
"learning_rate": 7.687154905648929e-06,
"loss": 0.0298,
"step": 406000
},
{
"epoch": 24.66,
"learning_rate": 7.672016261149203e-06,
"loss": 0.0304,
"step": 406500
},
{
"epoch": 24.7,
"learning_rate": 7.656847278684546e-06,
"loss": 0.0292,
"step": 407000
},
{
"epoch": 24.73,
"learning_rate": 7.64170863418482e-06,
"loss": 0.0302,
"step": 407500
},
{
"epoch": 24.76,
"learning_rate": 7.626539651720163e-06,
"loss": 0.0308,
"step": 408000
},
{
"epoch": 24.79,
"learning_rate": 7.6113706692555065e-06,
"loss": 0.0296,
"step": 408500
},
{
"epoch": 24.82,
"learning_rate": 7.596201686790851e-06,
"loss": 0.0294,
"step": 409000
},
{
"epoch": 24.85,
"learning_rate": 7.581032704326194e-06,
"loss": 0.0297,
"step": 409500
},
{
"epoch": 24.88,
"learning_rate": 7.565894059826467e-06,
"loss": 0.0302,
"step": 410000
},
{
"epoch": 24.91,
"learning_rate": 7.550725077361812e-06,
"loss": 0.0288,
"step": 410500
},
{
"epoch": 24.94,
"learning_rate": 7.535556094897155e-06,
"loss": 0.0303,
"step": 411000
},
{
"epoch": 24.97,
"learning_rate": 7.520387112432499e-06,
"loss": 0.03,
"step": 411500
},
{
"epoch": 25.0,
"learning_rate": 7.5052181299678425e-06,
"loss": 0.0293,
"step": 412000
},
{
"epoch": 25.0,
"eval_bleu": 87.6001,
"eval_gen_len": 14.8399,
"eval_loss": 0.2072879821062088,
"eval_runtime": 218.5505,
"eval_samples_per_second": 73.777,
"eval_steps_per_second": 2.306,
"step": 412025
},
{
"epoch": 25.03,
"learning_rate": 7.490079485468116e-06,
"loss": 0.0256,
"step": 412500
},
{
"epoch": 25.06,
"learning_rate": 7.47491050300346e-06,
"loss": 0.0272,
"step": 413000
},
{
"epoch": 25.09,
"learning_rate": 7.4597415205388025e-06,
"loss": 0.0268,
"step": 413500
},
{
"epoch": 25.12,
"learning_rate": 7.444572538074147e-06,
"loss": 0.0263,
"step": 414000
},
{
"epoch": 25.15,
"learning_rate": 7.42940355560949e-06,
"loss": 0.0268,
"step": 414500
},
{
"epoch": 25.18,
"learning_rate": 7.414234573144834e-06,
"loss": 0.0277,
"step": 415000
},
{
"epoch": 25.21,
"learning_rate": 7.3990655906801775e-06,
"loss": 0.0261,
"step": 415500
},
{
"epoch": 25.24,
"learning_rate": 7.383896608215522e-06,
"loss": 0.0269,
"step": 416000
},
{
"epoch": 25.27,
"learning_rate": 7.368757963715794e-06,
"loss": 0.0267,
"step": 416500
},
{
"epoch": 25.3,
"learning_rate": 7.3535889812511384e-06,
"loss": 0.0281,
"step": 417000
},
{
"epoch": 25.33,
"learning_rate": 7.338419998786483e-06,
"loss": 0.0279,
"step": 417500
},
{
"epoch": 25.36,
"learning_rate": 7.323281354286755e-06,
"loss": 0.0275,
"step": 418000
},
{
"epoch": 25.39,
"learning_rate": 7.308112371822099e-06,
"loss": 0.0264,
"step": 418500
},
{
"epoch": 25.42,
"learning_rate": 7.292943389357442e-06,
"loss": 0.0279,
"step": 419000
},
{
"epoch": 25.45,
"learning_rate": 7.277774406892786e-06,
"loss": 0.0276,
"step": 419500
},
{
"epoch": 25.48,
"learning_rate": 7.262605424428129e-06,
"loss": 0.0277,
"step": 420000
},
{
"epoch": 25.51,
"learning_rate": 7.2474364419634735e-06,
"loss": 0.0266,
"step": 420500
},
{
"epoch": 25.54,
"learning_rate": 7.232267459498818e-06,
"loss": 0.0277,
"step": 421000
},
{
"epoch": 25.57,
"learning_rate": 7.217098477034161e-06,
"loss": 0.0282,
"step": 421500
},
{
"epoch": 25.61,
"learning_rate": 7.201959832534434e-06,
"loss": 0.0269,
"step": 422000
},
{
"epoch": 25.64,
"learning_rate": 7.186790850069778e-06,
"loss": 0.0274,
"step": 422500
},
{
"epoch": 25.67,
"learning_rate": 7.171652205570051e-06,
"loss": 0.0279,
"step": 423000
},
{
"epoch": 25.7,
"learning_rate": 7.156483223105395e-06,
"loss": 0.028,
"step": 423500
},
{
"epoch": 25.73,
"learning_rate": 7.141314240640739e-06,
"loss": 0.0262,
"step": 424000
},
{
"epoch": 25.76,
"learning_rate": 7.126175596141012e-06,
"loss": 0.0279,
"step": 424500
},
{
"epoch": 25.79,
"learning_rate": 7.111006613676355e-06,
"loss": 0.0274,
"step": 425000
},
{
"epoch": 25.82,
"learning_rate": 7.0958376312116995e-06,
"loss": 0.0277,
"step": 425500
},
{
"epoch": 25.85,
"learning_rate": 7.080668648747042e-06,
"loss": 0.028,
"step": 426000
},
{
"epoch": 25.88,
"learning_rate": 7.065499666282386e-06,
"loss": 0.0278,
"step": 426500
},
{
"epoch": 25.91,
"learning_rate": 7.0503306838177295e-06,
"loss": 0.0282,
"step": 427000
},
{
"epoch": 25.94,
"learning_rate": 7.035161701353074e-06,
"loss": 0.0277,
"step": 427500
},
{
"epoch": 25.97,
"learning_rate": 7.019992718888418e-06,
"loss": 0.0287,
"step": 428000
},
{
"epoch": 26.0,
"learning_rate": 7.004823736423761e-06,
"loss": 0.0277,
"step": 428500
},
{
"epoch": 26.0,
"eval_bleu": 87.7954,
"eval_gen_len": 14.789,
"eval_loss": 0.2085237354040146,
"eval_runtime": 217.0143,
"eval_samples_per_second": 74.299,
"eval_steps_per_second": 2.322,
"step": 428506
},
{
"epoch": 26.03,
"learning_rate": 6.989685091924035e-06,
"loss": 0.025,
"step": 429000
},
{
"epoch": 26.06,
"learning_rate": 6.974516109459378e-06,
"loss": 0.0242,
"step": 429500
},
{
"epoch": 26.09,
"learning_rate": 6.959347126994722e-06,
"loss": 0.0256,
"step": 430000
},
{
"epoch": 26.12,
"learning_rate": 6.944178144530066e-06,
"loss": 0.0255,
"step": 430500
},
{
"epoch": 26.15,
"learning_rate": 6.929039500030339e-06,
"loss": 0.0246,
"step": 431000
},
{
"epoch": 26.18,
"learning_rate": 6.913870517565682e-06,
"loss": 0.0251,
"step": 431500
},
{
"epoch": 26.21,
"learning_rate": 6.8987015351010255e-06,
"loss": 0.0253,
"step": 432000
},
{
"epoch": 26.24,
"learning_rate": 6.88353255263637e-06,
"loss": 0.0257,
"step": 432500
},
{
"epoch": 26.27,
"learning_rate": 6.868363570171713e-06,
"loss": 0.0255,
"step": 433000
},
{
"epoch": 26.3,
"learning_rate": 6.853194587707057e-06,
"loss": 0.0255,
"step": 433500
},
{
"epoch": 26.33,
"learning_rate": 6.838025605242401e-06,
"loss": 0.0253,
"step": 434000
},
{
"epoch": 26.36,
"learning_rate": 6.822856622777745e-06,
"loss": 0.0256,
"step": 434500
},
{
"epoch": 26.39,
"learning_rate": 6.807717978278018e-06,
"loss": 0.0251,
"step": 435000
},
{
"epoch": 26.42,
"learning_rate": 6.792548995813361e-06,
"loss": 0.0248,
"step": 435500
},
{
"epoch": 26.45,
"learning_rate": 6.777380013348706e-06,
"loss": 0.0256,
"step": 436000
},
{
"epoch": 26.49,
"learning_rate": 6.76221103088405e-06,
"loss": 0.0256,
"step": 436500
},
{
"epoch": 26.52,
"learning_rate": 6.747042048419392e-06,
"loss": 0.0257,
"step": 437000
},
{
"epoch": 26.55,
"learning_rate": 6.731873065954736e-06,
"loss": 0.0257,
"step": 437500
},
{
"epoch": 26.58,
"learning_rate": 6.716734421455009e-06,
"loss": 0.025,
"step": 438000
},
{
"epoch": 26.61,
"learning_rate": 6.701565438990353e-06,
"loss": 0.0265,
"step": 438500
},
{
"epoch": 26.64,
"learning_rate": 6.6863964565256965e-06,
"loss": 0.0257,
"step": 439000
},
{
"epoch": 26.67,
"learning_rate": 6.671227474061041e-06,
"loss": 0.026,
"step": 439500
},
{
"epoch": 26.7,
"learning_rate": 6.656088829561313e-06,
"loss": 0.0249,
"step": 440000
},
{
"epoch": 26.73,
"learning_rate": 6.640919847096657e-06,
"loss": 0.0249,
"step": 440500
},
{
"epoch": 26.76,
"learning_rate": 6.6257508646320016e-06,
"loss": 0.0264,
"step": 441000
},
{
"epoch": 26.79,
"learning_rate": 6.610581882167345e-06,
"loss": 0.0254,
"step": 441500
},
{
"epoch": 26.82,
"learning_rate": 6.595412899702689e-06,
"loss": 0.0263,
"step": 442000
},
{
"epoch": 26.85,
"learning_rate": 6.580243917238032e-06,
"loss": 0.0263,
"step": 442500
},
{
"epoch": 26.88,
"learning_rate": 6.565074934773376e-06,
"loss": 0.0259,
"step": 443000
},
{
"epoch": 26.91,
"learning_rate": 6.549936290273648e-06,
"loss": 0.0257,
"step": 443500
},
{
"epoch": 26.94,
"learning_rate": 6.5347673078089925e-06,
"loss": 0.0264,
"step": 444000
},
{
"epoch": 26.97,
"learning_rate": 6.519598325344337e-06,
"loss": 0.0266,
"step": 444500
},
{
"epoch": 27.0,
"eval_bleu": 87.6557,
"eval_gen_len": 14.8303,
"eval_loss": 0.21119491755962372,
"eval_runtime": 217.2258,
"eval_samples_per_second": 74.227,
"eval_steps_per_second": 2.32,
"step": 444987
},
{
"epoch": 27.0,
"learning_rate": 6.50442934287968e-06,
"loss": 0.0266,
"step": 445000
},
{
"epoch": 27.03,
"learning_rate": 6.489260360415024e-06,
"loss": 0.0232,
"step": 445500
},
{
"epoch": 27.06,
"learning_rate": 6.4740913779503675e-06,
"loss": 0.0232,
"step": 446000
},
{
"epoch": 27.09,
"learning_rate": 6.458952733450641e-06,
"loss": 0.0232,
"step": 446500
},
{
"epoch": 27.12,
"learning_rate": 6.443783750985984e-06,
"loss": 0.0231,
"step": 447000
},
{
"epoch": 27.15,
"learning_rate": 6.4286451064862576e-06,
"loss": 0.0233,
"step": 447500
},
{
"epoch": 27.18,
"learning_rate": 6.413476124021602e-06,
"loss": 0.0233,
"step": 448000
},
{
"epoch": 27.21,
"learning_rate": 6.398307141556945e-06,
"loss": 0.0235,
"step": 448500
},
{
"epoch": 27.24,
"learning_rate": 6.3831381590922884e-06,
"loss": 0.024,
"step": 449000
},
{
"epoch": 27.27,
"learning_rate": 6.367969176627632e-06,
"loss": 0.0238,
"step": 449500
},
{
"epoch": 27.3,
"learning_rate": 6.352800194162976e-06,
"loss": 0.0239,
"step": 450000
},
{
"epoch": 27.33,
"learning_rate": 6.337631211698319e-06,
"loss": 0.0237,
"step": 450500
},
{
"epoch": 27.36,
"learning_rate": 6.3224622292336635e-06,
"loss": 0.0233,
"step": 451000
},
{
"epoch": 27.4,
"learning_rate": 6.307293246769008e-06,
"loss": 0.024,
"step": 451500
},
{
"epoch": 27.43,
"learning_rate": 6.292124264304351e-06,
"loss": 0.024,
"step": 452000
},
{
"epoch": 27.46,
"learning_rate": 6.276955281839695e-06,
"loss": 0.0241,
"step": 452500
},
{
"epoch": 27.49,
"learning_rate": 6.2617862993750385e-06,
"loss": 0.0244,
"step": 453000
},
{
"epoch": 27.52,
"learning_rate": 6.246617316910382e-06,
"loss": 0.0244,
"step": 453500
},
{
"epoch": 27.55,
"learning_rate": 6.231478672410656e-06,
"loss": 0.0243,
"step": 454000
},
{
"epoch": 27.58,
"learning_rate": 6.216340027910929e-06,
"loss": 0.0242,
"step": 454500
},
{
"epoch": 27.61,
"learning_rate": 6.201171045446271e-06,
"loss": 0.0228,
"step": 455000
},
{
"epoch": 27.64,
"learning_rate": 6.186002062981615e-06,
"loss": 0.0243,
"step": 455500
},
{
"epoch": 27.67,
"learning_rate": 6.1708330805169595e-06,
"loss": 0.0242,
"step": 456000
},
{
"epoch": 27.7,
"learning_rate": 6.155664098052303e-06,
"loss": 0.0238,
"step": 456500
},
{
"epoch": 27.73,
"learning_rate": 6.140495115587647e-06,
"loss": 0.0237,
"step": 457000
},
{
"epoch": 27.76,
"learning_rate": 6.1253564710879195e-06,
"loss": 0.0255,
"step": 457500
},
{
"epoch": 27.79,
"learning_rate": 6.110217826588193e-06,
"loss": 0.025,
"step": 458000
},
{
"epoch": 27.82,
"learning_rate": 6.095048844123537e-06,
"loss": 0.0244,
"step": 458500
},
{
"epoch": 27.85,
"learning_rate": 6.07987986165888e-06,
"loss": 0.0248,
"step": 459000
},
{
"epoch": 27.88,
"learning_rate": 6.0647108791942246e-06,
"loss": 0.0251,
"step": 459500
},
{
"epoch": 27.91,
"learning_rate": 6.049541896729568e-06,
"loss": 0.0249,
"step": 460000
},
{
"epoch": 27.94,
"learning_rate": 6.034372914264912e-06,
"loss": 0.0251,
"step": 460500
},
{
"epoch": 27.97,
"learning_rate": 6.0192039318002546e-06,
"loss": 0.0247,
"step": 461000
},
{
"epoch": 28.0,
"eval_bleu": 87.8108,
"eval_gen_len": 14.797,
"eval_loss": 0.21366077661514282,
"eval_runtime": 216.9055,
"eval_samples_per_second": 74.337,
"eval_steps_per_second": 2.324,
"step": 461468
},
{
"epoch": 28.0,
"learning_rate": 6.004034949335599e-06,
"loss": 0.0241,
"step": 461500
},
{
"epoch": 28.03,
"learning_rate": 5.988865966870943e-06,
"loss": 0.0218,
"step": 462000
},
{
"epoch": 28.06,
"learning_rate": 5.973696984406286e-06,
"loss": 0.0221,
"step": 462500
},
{
"epoch": 28.09,
"learning_rate": 5.95855833990656e-06,
"loss": 0.0213,
"step": 463000
},
{
"epoch": 28.12,
"learning_rate": 5.943389357441903e-06,
"loss": 0.0219,
"step": 463500
},
{
"epoch": 28.15,
"learning_rate": 5.928220374977247e-06,
"loss": 0.0223,
"step": 464000
},
{
"epoch": 28.18,
"learning_rate": 5.913051392512591e-06,
"loss": 0.0222,
"step": 464500
},
{
"epoch": 28.21,
"learning_rate": 5.897882410047935e-06,
"loss": 0.0222,
"step": 465000
},
{
"epoch": 28.24,
"learning_rate": 5.882743765548208e-06,
"loss": 0.0226,
"step": 465500
},
{
"epoch": 28.27,
"learning_rate": 5.867574783083551e-06,
"loss": 0.0224,
"step": 466000
},
{
"epoch": 28.31,
"learning_rate": 5.8524058006188956e-06,
"loss": 0.0224,
"step": 466500
},
{
"epoch": 28.34,
"learning_rate": 5.837236818154238e-06,
"loss": 0.0229,
"step": 467000
},
{
"epoch": 28.37,
"learning_rate": 5.822098173654511e-06,
"loss": 0.0222,
"step": 467500
},
{
"epoch": 28.4,
"learning_rate": 5.806929191189855e-06,
"loss": 0.0232,
"step": 468000
},
{
"epoch": 28.43,
"learning_rate": 5.791760208725199e-06,
"loss": 0.0227,
"step": 468500
},
{
"epoch": 28.46,
"learning_rate": 5.776591226260543e-06,
"loss": 0.0231,
"step": 469000
},
{
"epoch": 28.49,
"learning_rate": 5.761452581760816e-06,
"loss": 0.0218,
"step": 469500
},
{
"epoch": 28.52,
"learning_rate": 5.74628359929616e-06,
"loss": 0.0228,
"step": 470000
},
{
"epoch": 28.55,
"learning_rate": 5.731114616831503e-06,
"loss": 0.0228,
"step": 470500
},
{
"epoch": 28.58,
"learning_rate": 5.715945634366847e-06,
"loss": 0.023,
"step": 471000
},
{
"epoch": 28.61,
"learning_rate": 5.7007766519021915e-06,
"loss": 0.0224,
"step": 471500
},
{
"epoch": 28.64,
"learning_rate": 5.685607669437535e-06,
"loss": 0.0225,
"step": 472000
},
{
"epoch": 28.67,
"learning_rate": 5.670438686972878e-06,
"loss": 0.0227,
"step": 472500
},
{
"epoch": 28.7,
"learning_rate": 5.6552697045082216e-06,
"loss": 0.0232,
"step": 473000
},
{
"epoch": 28.73,
"learning_rate": 5.640100722043566e-06,
"loss": 0.0237,
"step": 473500
},
{
"epoch": 28.76,
"learning_rate": 5.624931739578909e-06,
"loss": 0.0222,
"step": 474000
},
{
"epoch": 28.79,
"learning_rate": 5.6097930950791824e-06,
"loss": 0.0224,
"step": 474500
},
{
"epoch": 28.82,
"learning_rate": 5.594624112614527e-06,
"loss": 0.0226,
"step": 475000
},
{
"epoch": 28.85,
"learning_rate": 5.57945513014987e-06,
"loss": 0.0223,
"step": 475500
},
{
"epoch": 28.88,
"learning_rate": 5.564286147685214e-06,
"loss": 0.0227,
"step": 476000
},
{
"epoch": 28.91,
"learning_rate": 5.5491171652205575e-06,
"loss": 0.0225,
"step": 476500
},
{
"epoch": 28.94,
"learning_rate": 5.533948182755902e-06,
"loss": 0.0226,
"step": 477000
},
{
"epoch": 28.97,
"learning_rate": 5.518779200291244e-06,
"loss": 0.0232,
"step": 477500
},
{
"epoch": 29.0,
"eval_bleu": 87.6826,
"eval_gen_len": 14.7702,
"eval_loss": 0.21583835780620575,
"eval_runtime": 217.0792,
"eval_samples_per_second": 74.277,
"eval_steps_per_second": 2.322,
"step": 477949
},
{
"epoch": 29.0,
"learning_rate": 5.503640555791518e-06,
"loss": 0.0228,
"step": 478000
},
{
"epoch": 29.03,
"learning_rate": 5.488471573326861e-06,
"loss": 0.0209,
"step": 478500
},
{
"epoch": 29.06,
"learning_rate": 5.473302590862205e-06,
"loss": 0.0209,
"step": 479000
},
{
"epoch": 29.09,
"learning_rate": 5.458133608397549e-06,
"loss": 0.021,
"step": 479500
},
{
"epoch": 29.12,
"learning_rate": 5.4429646259328926e-06,
"loss": 0.0199,
"step": 480000
},
{
"epoch": 29.15,
"learning_rate": 5.427795643468237e-06,
"loss": 0.0207,
"step": 480500
},
{
"epoch": 29.19,
"learning_rate": 5.412656998968509e-06,
"loss": 0.0213,
"step": 481000
},
{
"epoch": 29.22,
"learning_rate": 5.3974880165038535e-06,
"loss": 0.0207,
"step": 481500
},
{
"epoch": 29.25,
"learning_rate": 5.382319034039198e-06,
"loss": 0.0212,
"step": 482000
},
{
"epoch": 29.28,
"learning_rate": 5.367150051574541e-06,
"loss": 0.0211,
"step": 482500
},
{
"epoch": 29.31,
"learning_rate": 5.351981069109885e-06,
"loss": 0.0218,
"step": 483000
},
{
"epoch": 29.34,
"learning_rate": 5.336842424610158e-06,
"loss": 0.0207,
"step": 483500
},
{
"epoch": 29.37,
"learning_rate": 5.321673442145502e-06,
"loss": 0.0205,
"step": 484000
},
{
"epoch": 29.4,
"learning_rate": 5.306504459680844e-06,
"loss": 0.0205,
"step": 484500
},
{
"epoch": 29.43,
"learning_rate": 5.2913354772161885e-06,
"loss": 0.0213,
"step": 485000
},
{
"epoch": 29.46,
"learning_rate": 5.276166494751533e-06,
"loss": 0.022,
"step": 485500
},
{
"epoch": 29.49,
"learning_rate": 5.260997512286876e-06,
"loss": 0.0212,
"step": 486000
},
{
"epoch": 29.52,
"learning_rate": 5.24582852982222e-06,
"loss": 0.0213,
"step": 486500
},
{
"epoch": 29.55,
"learning_rate": 5.230659547357564e-06,
"loss": 0.021,
"step": 487000
},
{
"epoch": 29.58,
"learning_rate": 5.215490564892908e-06,
"loss": 0.0213,
"step": 487500
},
{
"epoch": 29.61,
"learning_rate": 5.20035192039318e-06,
"loss": 0.0212,
"step": 488000
},
{
"epoch": 29.64,
"learning_rate": 5.1851829379285245e-06,
"loss": 0.0213,
"step": 488500
},
{
"epoch": 29.67,
"learning_rate": 5.170013955463869e-06,
"loss": 0.0216,
"step": 489000
},
{
"epoch": 29.7,
"learning_rate": 5.154844972999211e-06,
"loss": 0.0215,
"step": 489500
},
{
"epoch": 29.73,
"learning_rate": 5.139675990534555e-06,
"loss": 0.0211,
"step": 490000
},
{
"epoch": 29.76,
"learning_rate": 5.124537346034828e-06,
"loss": 0.0213,
"step": 490500
},
{
"epoch": 29.79,
"learning_rate": 5.109398701535101e-06,
"loss": 0.0213,
"step": 491000
},
{
"epoch": 29.82,
"learning_rate": 5.0942297190704445e-06,
"loss": 0.0207,
"step": 491500
},
{
"epoch": 29.85,
"learning_rate": 5.079060736605789e-06,
"loss": 0.0217,
"step": 492000
},
{
"epoch": 29.88,
"learning_rate": 5.063891754141133e-06,
"loss": 0.0222,
"step": 492500
},
{
"epoch": 29.91,
"learning_rate": 5.048783447606335e-06,
"loss": 0.0215,
"step": 493000
},
{
"epoch": 29.94,
"learning_rate": 5.033614465141679e-06,
"loss": 0.022,
"step": 493500
},
{
"epoch": 29.97,
"learning_rate": 5.018445482677022e-06,
"loss": 0.0225,
"step": 494000
},
{
"epoch": 30.0,
"eval_bleu": 87.759,
"eval_gen_len": 14.7979,
"eval_loss": 0.21940629184246063,
"eval_runtime": 217.8188,
"eval_samples_per_second": 74.025,
"eval_steps_per_second": 2.314,
"step": 494430
},
{
"epoch": 30.0,
"learning_rate": 5.003276500212366e-06,
"loss": 0.0203,
"step": 494500
},
{
"epoch": 30.03,
"learning_rate": 4.98810751774771e-06,
"loss": 0.0189,
"step": 495000
},
{
"epoch": 30.06,
"learning_rate": 4.972938535283053e-06,
"loss": 0.0198,
"step": 495500
},
{
"epoch": 30.1,
"learning_rate": 4.957769552818397e-06,
"loss": 0.0201,
"step": 496000
},
{
"epoch": 30.13,
"learning_rate": 4.942600570353741e-06,
"loss": 0.0194,
"step": 496500
},
{
"epoch": 30.16,
"learning_rate": 4.927431587889085e-06,
"loss": 0.0199,
"step": 497000
},
{
"epoch": 30.19,
"learning_rate": 4.912262605424429e-06,
"loss": 0.02,
"step": 497500
},
{
"epoch": 30.22,
"learning_rate": 4.897093622959772e-06,
"loss": 0.0194,
"step": 498000
},
{
"epoch": 30.25,
"learning_rate": 4.8819246404951156e-06,
"loss": 0.0199,
"step": 498500
},
{
"epoch": 30.28,
"learning_rate": 4.86675565803046e-06,
"loss": 0.02,
"step": 499000
},
{
"epoch": 30.31,
"learning_rate": 4.851586675565804e-06,
"loss": 0.0198,
"step": 499500
},
{
"epoch": 30.34,
"learning_rate": 4.836417693101147e-06,
"loss": 0.0195,
"step": 500000
},
{
"epoch": 30.37,
"learning_rate": 4.821248710636491e-06,
"loss": 0.0194,
"step": 500500
},
{
"epoch": 30.4,
"learning_rate": 4.806079728171835e-06,
"loss": 0.0192,
"step": 501000
},
{
"epoch": 30.43,
"learning_rate": 4.790941083672107e-06,
"loss": 0.0193,
"step": 501500
},
{
"epoch": 30.46,
"learning_rate": 4.775802439172381e-06,
"loss": 0.0201,
"step": 502000
},
{
"epoch": 30.49,
"learning_rate": 4.760633456707725e-06,
"loss": 0.0201,
"step": 502500
},
{
"epoch": 30.52,
"learning_rate": 4.745464474243068e-06,
"loss": 0.0209,
"step": 503000
},
{
"epoch": 30.55,
"learning_rate": 4.7302954917784115e-06,
"loss": 0.0202,
"step": 503500
},
{
"epoch": 30.58,
"learning_rate": 4.715156847278685e-06,
"loss": 0.0209,
"step": 504000
},
{
"epoch": 30.61,
"learning_rate": 4.699987864814028e-06,
"loss": 0.02,
"step": 504500
},
{
"epoch": 30.64,
"learning_rate": 4.684818882349372e-06,
"loss": 0.0206,
"step": 505000
},
{
"epoch": 30.67,
"learning_rate": 4.669649899884717e-06,
"loss": 0.02,
"step": 505500
},
{
"epoch": 30.7,
"learning_rate": 4.65448091742006e-06,
"loss": 0.0211,
"step": 506000
},
{
"epoch": 30.73,
"learning_rate": 4.639311934955403e-06,
"loss": 0.0203,
"step": 506500
},
{
"epoch": 30.76,
"learning_rate": 4.6241429524907475e-06,
"loss": 0.0202,
"step": 507000
},
{
"epoch": 30.79,
"learning_rate": 4.608973970026091e-06,
"loss": 0.0207,
"step": 507500
},
{
"epoch": 30.82,
"learning_rate": 4.593804987561435e-06,
"loss": 0.0211,
"step": 508000
},
{
"epoch": 30.85,
"learning_rate": 4.5786663430617075e-06,
"loss": 0.0206,
"step": 508500
},
{
"epoch": 30.88,
"learning_rate": 4.563497360597052e-06,
"loss": 0.0204,
"step": 509000
},
{
"epoch": 30.91,
"learning_rate": 4.548328378132395e-06,
"loss": 0.0204,
"step": 509500
},
{
"epoch": 30.94,
"learning_rate": 4.533159395667739e-06,
"loss": 0.0195,
"step": 510000
},
{
"epoch": 30.98,
"learning_rate": 4.5179904132030825e-06,
"loss": 0.0205,
"step": 510500
},
{
"epoch": 31.0,
"eval_bleu": 87.805,
"eval_gen_len": 14.8214,
"eval_loss": 0.22064241766929626,
"eval_runtime": 221.1914,
"eval_samples_per_second": 72.896,
"eval_steps_per_second": 2.279,
"step": 510911
},
{
"epoch": 31.01,
"learning_rate": 4.502821430738427e-06,
"loss": 0.0203,
"step": 511000
},
{
"epoch": 31.04,
"learning_rate": 4.487682786238699e-06,
"loss": 0.0182,
"step": 511500
},
{
"epoch": 31.07,
"learning_rate": 4.4725138037740434e-06,
"loss": 0.0175,
"step": 512000
},
{
"epoch": 31.1,
"learning_rate": 4.457344821309387e-06,
"loss": 0.0187,
"step": 512500
},
{
"epoch": 31.13,
"learning_rate": 4.442175838844731e-06,
"loss": 0.0184,
"step": 513000
},
{
"epoch": 31.16,
"learning_rate": 4.427006856380074e-06,
"loss": 0.0183,
"step": 513500
},
{
"epoch": 31.19,
"learning_rate": 4.411868211880348e-06,
"loss": 0.019,
"step": 514000
},
{
"epoch": 31.22,
"learning_rate": 4.396699229415691e-06,
"loss": 0.0191,
"step": 514500
},
{
"epoch": 31.25,
"learning_rate": 4.381530246951035e-06,
"loss": 0.0189,
"step": 515000
},
{
"epoch": 31.28,
"learning_rate": 4.3663916024513085e-06,
"loss": 0.018,
"step": 515500
},
{
"epoch": 31.31,
"learning_rate": 4.351222619986651e-06,
"loss": 0.0192,
"step": 516000
},
{
"epoch": 31.34,
"learning_rate": 4.336053637521995e-06,
"loss": 0.0195,
"step": 516500
},
{
"epoch": 31.37,
"learning_rate": 4.320884655057339e-06,
"loss": 0.0189,
"step": 517000
},
{
"epoch": 31.4,
"learning_rate": 4.305715672592683e-06,
"loss": 0.0186,
"step": 517500
},
{
"epoch": 31.43,
"learning_rate": 4.290577028092956e-06,
"loss": 0.0186,
"step": 518000
},
{
"epoch": 31.46,
"learning_rate": 4.2754080456282994e-06,
"loss": 0.0196,
"step": 518500
},
{
"epoch": 31.49,
"learning_rate": 4.260239063163643e-06,
"loss": 0.0197,
"step": 519000
},
{
"epoch": 31.52,
"learning_rate": 4.245070080698987e-06,
"loss": 0.0191,
"step": 519500
},
{
"epoch": 31.55,
"learning_rate": 4.229901098234331e-06,
"loss": 0.0191,
"step": 520000
},
{
"epoch": 31.58,
"learning_rate": 4.2147321157696745e-06,
"loss": 0.0195,
"step": 520500
},
{
"epoch": 31.61,
"learning_rate": 4.199563133305018e-06,
"loss": 0.0193,
"step": 521000
},
{
"epoch": 31.64,
"learning_rate": 4.184394150840362e-06,
"loss": 0.0183,
"step": 521500
},
{
"epoch": 31.67,
"learning_rate": 4.169225168375705e-06,
"loss": 0.0195,
"step": 522000
},
{
"epoch": 31.7,
"learning_rate": 4.1540561859110495e-06,
"loss": 0.0186,
"step": 522500
},
{
"epoch": 31.73,
"learning_rate": 4.138887203446394e-06,
"loss": 0.0185,
"step": 523000
},
{
"epoch": 31.76,
"learning_rate": 4.123748558946666e-06,
"loss": 0.0189,
"step": 523500
},
{
"epoch": 31.79,
"learning_rate": 4.1085795764820096e-06,
"loss": 0.0194,
"step": 524000
},
{
"epoch": 31.82,
"learning_rate": 4.093410594017354e-06,
"loss": 0.0198,
"step": 524500
},
{
"epoch": 31.85,
"learning_rate": 4.078241611552697e-06,
"loss": 0.0197,
"step": 525000
},
{
"epoch": 31.89,
"learning_rate": 4.0631029670529704e-06,
"loss": 0.0193,
"step": 525500
},
{
"epoch": 31.92,
"learning_rate": 4.047933984588315e-06,
"loss": 0.0197,
"step": 526000
},
{
"epoch": 31.95,
"learning_rate": 4.032795340088587e-06,
"loss": 0.0202,
"step": 526500
},
{
"epoch": 31.98,
"learning_rate": 4.017626357623931e-06,
"loss": 0.0191,
"step": 527000
},
{
"epoch": 32.0,
"eval_bleu": 87.9338,
"eval_gen_len": 14.778,
"eval_loss": 0.22409668564796448,
"eval_runtime": 218.5886,
"eval_samples_per_second": 73.764,
"eval_steps_per_second": 2.306,
"step": 527392
},
{
"epoch": 32.01,
"learning_rate": 4.002457375159275e-06,
"loss": 0.019,
"step": 527500
},
{
"epoch": 32.04,
"learning_rate": 3.987288392694618e-06,
"loss": 0.0174,
"step": 528000
},
{
"epoch": 32.07,
"learning_rate": 3.972119410229962e-06,
"loss": 0.0174,
"step": 528500
},
{
"epoch": 32.1,
"learning_rate": 3.9569504277653055e-06,
"loss": 0.0175,
"step": 529000
},
{
"epoch": 32.13,
"learning_rate": 3.94178144530065e-06,
"loss": 0.0179,
"step": 529500
},
{
"epoch": 32.16,
"learning_rate": 3.926612462835993e-06,
"loss": 0.0178,
"step": 530000
},
{
"epoch": 32.19,
"learning_rate": 3.911443480371337e-06,
"loss": 0.0178,
"step": 530500
},
{
"epoch": 32.22,
"learning_rate": 3.89630483587161e-06,
"loss": 0.0185,
"step": 531000
},
{
"epoch": 32.25,
"learning_rate": 3.881135853406954e-06,
"loss": 0.0188,
"step": 531500
},
{
"epoch": 32.28,
"learning_rate": 3.865966870942297e-06,
"loss": 0.0178,
"step": 532000
},
{
"epoch": 32.31,
"learning_rate": 3.8507978884776415e-06,
"loss": 0.0183,
"step": 532500
},
{
"epoch": 32.34,
"learning_rate": 3.835628906012985e-06,
"loss": 0.0178,
"step": 533000
},
{
"epoch": 32.37,
"learning_rate": 3.820490261513258e-06,
"loss": 0.0178,
"step": 533500
},
{
"epoch": 32.4,
"learning_rate": 3.8053212790486015e-06,
"loss": 0.0184,
"step": 534000
},
{
"epoch": 32.43,
"learning_rate": 3.7901522965839453e-06,
"loss": 0.0183,
"step": 534500
},
{
"epoch": 32.46,
"learning_rate": 3.7750136520842186e-06,
"loss": 0.0181,
"step": 535000
},
{
"epoch": 32.49,
"learning_rate": 3.7598446696195624e-06,
"loss": 0.0178,
"step": 535500
},
{
"epoch": 32.52,
"learning_rate": 3.744675687154906e-06,
"loss": 0.0182,
"step": 536000
},
{
"epoch": 32.55,
"learning_rate": 3.7295067046902495e-06,
"loss": 0.0185,
"step": 536500
},
{
"epoch": 32.58,
"learning_rate": 3.7143377222255932e-06,
"loss": 0.0182,
"step": 537000
},
{
"epoch": 32.61,
"learning_rate": 3.699168739760937e-06,
"loss": 0.0185,
"step": 537500
},
{
"epoch": 32.64,
"learning_rate": 3.683999757296281e-06,
"loss": 0.0183,
"step": 538000
},
{
"epoch": 32.67,
"learning_rate": 3.668830774831625e-06,
"loss": 0.0176,
"step": 538500
},
{
"epoch": 32.7,
"learning_rate": 3.6536617923669683e-06,
"loss": 0.0186,
"step": 539000
},
{
"epoch": 32.73,
"learning_rate": 3.638492809902312e-06,
"loss": 0.0181,
"step": 539500
},
{
"epoch": 32.77,
"learning_rate": 3.623354165402585e-06,
"loss": 0.018,
"step": 540000
},
{
"epoch": 32.8,
"learning_rate": 3.6081851829379288e-06,
"loss": 0.0183,
"step": 540500
},
{
"epoch": 32.83,
"learning_rate": 3.5930162004732725e-06,
"loss": 0.0184,
"step": 541000
},
{
"epoch": 32.86,
"learning_rate": 3.577847218008616e-06,
"loss": 0.0182,
"step": 541500
},
{
"epoch": 32.89,
"learning_rate": 3.5627085735088896e-06,
"loss": 0.0181,
"step": 542000
},
{
"epoch": 32.92,
"learning_rate": 3.5475699290091626e-06,
"loss": 0.0187,
"step": 542500
},
{
"epoch": 32.95,
"learning_rate": 3.532400946544506e-06,
"loss": 0.0191,
"step": 543000
},
{
"epoch": 32.98,
"learning_rate": 3.5172319640798497e-06,
"loss": 0.0185,
"step": 543500
},
{
"epoch": 33.0,
"eval_bleu": 87.8214,
"eval_gen_len": 14.792,
"eval_loss": 0.22567500174045563,
"eval_runtime": 224.8959,
"eval_samples_per_second": 71.695,
"eval_steps_per_second": 2.241,
"step": 543873
},
{
"epoch": 33.01,
"learning_rate": 3.5020629816151934e-06,
"loss": 0.0179,
"step": 544000
},
{
"epoch": 33.04,
"learning_rate": 3.486893999150537e-06,
"loss": 0.0169,
"step": 544500
},
{
"epoch": 33.07,
"learning_rate": 3.4717250166858814e-06,
"loss": 0.0163,
"step": 545000
},
{
"epoch": 33.1,
"learning_rate": 3.4565560342212247e-06,
"loss": 0.0168,
"step": 545500
},
{
"epoch": 33.13,
"learning_rate": 3.4414173897214977e-06,
"loss": 0.0174,
"step": 546000
},
{
"epoch": 33.16,
"learning_rate": 3.4262484072568414e-06,
"loss": 0.017,
"step": 546500
},
{
"epoch": 33.19,
"learning_rate": 3.411079424792185e-06,
"loss": 0.0174,
"step": 547000
},
{
"epoch": 33.22,
"learning_rate": 3.395910442327529e-06,
"loss": 0.0174,
"step": 547500
},
{
"epoch": 33.25,
"learning_rate": 3.380741459862873e-06,
"loss": 0.0168,
"step": 548000
},
{
"epoch": 33.28,
"learning_rate": 3.365572477398216e-06,
"loss": 0.0176,
"step": 548500
},
{
"epoch": 33.31,
"learning_rate": 3.3504034949335602e-06,
"loss": 0.017,
"step": 549000
},
{
"epoch": 33.34,
"learning_rate": 3.335234512468904e-06,
"loss": 0.0174,
"step": 549500
},
{
"epoch": 33.37,
"learning_rate": 3.320095867969177e-06,
"loss": 0.0174,
"step": 550000
},
{
"epoch": 33.4,
"learning_rate": 3.3049268855045207e-06,
"loss": 0.0178,
"step": 550500
},
{
"epoch": 33.43,
"learning_rate": 3.289757903039864e-06,
"loss": 0.0174,
"step": 551000
},
{
"epoch": 33.46,
"learning_rate": 3.2745889205752078e-06,
"loss": 0.0171,
"step": 551500
},
{
"epoch": 33.49,
"learning_rate": 3.259419938110552e-06,
"loss": 0.0169,
"step": 552000
},
{
"epoch": 33.52,
"learning_rate": 3.2442509556458957e-06,
"loss": 0.017,
"step": 552500
},
{
"epoch": 33.55,
"learning_rate": 3.2291123111461687e-06,
"loss": 0.0187,
"step": 553000
},
{
"epoch": 33.58,
"learning_rate": 3.2139736666464416e-06,
"loss": 0.0176,
"step": 553500
},
{
"epoch": 33.61,
"learning_rate": 3.1988046841817854e-06,
"loss": 0.0174,
"step": 554000
},
{
"epoch": 33.64,
"learning_rate": 3.183635701717129e-06,
"loss": 0.0174,
"step": 554500
},
{
"epoch": 33.68,
"learning_rate": 3.1684667192524725e-06,
"loss": 0.0172,
"step": 555000
},
{
"epoch": 33.71,
"learning_rate": 3.1532977367878167e-06,
"loss": 0.0174,
"step": 555500
},
{
"epoch": 33.74,
"learning_rate": 3.1381287543231604e-06,
"loss": 0.017,
"step": 556000
},
{
"epoch": 33.77,
"learning_rate": 3.122959771858504e-06,
"loss": 0.0181,
"step": 556500
},
{
"epoch": 33.8,
"learning_rate": 3.1077907893938475e-06,
"loss": 0.0175,
"step": 557000
},
{
"epoch": 33.83,
"learning_rate": 3.0926218069291913e-06,
"loss": 0.0167,
"step": 557500
},
{
"epoch": 33.86,
"learning_rate": 3.0774831624294642e-06,
"loss": 0.018,
"step": 558000
},
{
"epoch": 33.89,
"learning_rate": 3.062314179964808e-06,
"loss": 0.0176,
"step": 558500
},
{
"epoch": 33.92,
"learning_rate": 3.047145197500152e-06,
"loss": 0.018,
"step": 559000
},
{
"epoch": 33.95,
"learning_rate": 3.031976215035496e-06,
"loss": 0.0174,
"step": 559500
},
{
"epoch": 33.98,
"learning_rate": 3.016837570535769e-06,
"loss": 0.0177,
"step": 560000
},
{
"epoch": 34.0,
"eval_bleu": 87.793,
"eval_gen_len": 14.7892,
"eval_loss": 0.22562597692012787,
"eval_runtime": 226.3249,
"eval_samples_per_second": 71.243,
"eval_steps_per_second": 2.227,
"step": 560354
},
{
"epoch": 34.01,
"learning_rate": 3.001668588071112e-06,
"loss": 0.0177,
"step": 560500
},
{
"epoch": 34.04,
"learning_rate": 2.986499605606456e-06,
"loss": 0.0161,
"step": 561000
},
{
"epoch": 34.07,
"learning_rate": 2.9713306231417997e-06,
"loss": 0.0154,
"step": 561500
},
{
"epoch": 34.1,
"learning_rate": 2.9561919786420727e-06,
"loss": 0.017,
"step": 562000
},
{
"epoch": 34.13,
"learning_rate": 2.941053334142346e-06,
"loss": 0.0161,
"step": 562500
},
{
"epoch": 34.16,
"learning_rate": 2.9258843516776898e-06,
"loss": 0.0164,
"step": 563000
},
{
"epoch": 34.19,
"learning_rate": 2.9107153692130335e-06,
"loss": 0.016,
"step": 563500
},
{
"epoch": 34.22,
"learning_rate": 2.8955463867483773e-06,
"loss": 0.0172,
"step": 564000
},
{
"epoch": 34.25,
"learning_rate": 2.8803774042837206e-06,
"loss": 0.0156,
"step": 564500
},
{
"epoch": 34.28,
"learning_rate": 2.8652084218190644e-06,
"loss": 0.0169,
"step": 565000
},
{
"epoch": 34.31,
"learning_rate": 2.8500394393544086e-06,
"loss": 0.0171,
"step": 565500
},
{
"epoch": 34.34,
"learning_rate": 2.8349007948546815e-06,
"loss": 0.0161,
"step": 566000
},
{
"epoch": 34.37,
"learning_rate": 2.8197318123900253e-06,
"loss": 0.0168,
"step": 566500
},
{
"epoch": 34.4,
"learning_rate": 2.804562829925369e-06,
"loss": 0.0164,
"step": 567000
},
{
"epoch": 34.43,
"learning_rate": 2.7893938474607124e-06,
"loss": 0.0166,
"step": 567500
},
{
"epoch": 34.46,
"learning_rate": 2.7742552029609853e-06,
"loss": 0.0164,
"step": 568000
},
{
"epoch": 34.49,
"learning_rate": 2.759086220496329e-06,
"loss": 0.0169,
"step": 568500
},
{
"epoch": 34.52,
"learning_rate": 2.7439172380316733e-06,
"loss": 0.0168,
"step": 569000
},
{
"epoch": 34.55,
"learning_rate": 2.728748255567017e-06,
"loss": 0.0164,
"step": 569500
},
{
"epoch": 34.59,
"learning_rate": 2.7135792731023604e-06,
"loss": 0.0164,
"step": 570000
},
{
"epoch": 34.62,
"learning_rate": 2.6984406286026337e-06,
"loss": 0.0166,
"step": 570500
},
{
"epoch": 34.65,
"learning_rate": 2.683271646137977e-06,
"loss": 0.0164,
"step": 571000
},
{
"epoch": 34.68,
"learning_rate": 2.668102663673321e-06,
"loss": 0.0167,
"step": 571500
},
{
"epoch": 34.71,
"learning_rate": 2.6529336812086646e-06,
"loss": 0.0172,
"step": 572000
},
{
"epoch": 34.74,
"learning_rate": 2.6377646987440088e-06,
"loss": 0.0172,
"step": 572500
},
{
"epoch": 34.77,
"learning_rate": 2.622595716279352e-06,
"loss": 0.0166,
"step": 573000
},
{
"epoch": 34.8,
"learning_rate": 2.607426733814696e-06,
"loss": 0.0166,
"step": 573500
},
{
"epoch": 34.83,
"learning_rate": 2.5922577513500396e-06,
"loss": 0.017,
"step": 574000
},
{
"epoch": 34.86,
"learning_rate": 2.5770887688853834e-06,
"loss": 0.0164,
"step": 574500
},
{
"epoch": 34.89,
"learning_rate": 2.5619501243856563e-06,
"loss": 0.0174,
"step": 575000
},
{
"epoch": 34.92,
"learning_rate": 2.5467811419210005e-06,
"loss": 0.0167,
"step": 575500
},
{
"epoch": 34.95,
"learning_rate": 2.5316424974212735e-06,
"loss": 0.0169,
"step": 576000
},
{
"epoch": 34.98,
"learning_rate": 2.5164735149566172e-06,
"loss": 0.0172,
"step": 576500
},
{
"epoch": 35.0,
"eval_bleu": 87.8156,
"eval_gen_len": 14.7718,
"eval_loss": 0.22844606637954712,
"eval_runtime": 225.457,
"eval_samples_per_second": 71.517,
"eval_steps_per_second": 2.235,
"step": 576835
},
{
"epoch": 35.01,
"learning_rate": 2.5013045324919606e-06,
"loss": 0.0161,
"step": 577000
},
{
"epoch": 35.04,
"learning_rate": 2.4861355500273043e-06,
"loss": 0.0152,
"step": 577500
},
{
"epoch": 35.07,
"learning_rate": 2.470966567562648e-06,
"loss": 0.0162,
"step": 578000
},
{
"epoch": 35.1,
"learning_rate": 2.455827923062921e-06,
"loss": 0.0156,
"step": 578500
},
{
"epoch": 35.13,
"learning_rate": 2.4406589405982648e-06,
"loss": 0.0154,
"step": 579000
},
{
"epoch": 35.16,
"learning_rate": 2.4254899581336085e-06,
"loss": 0.0161,
"step": 579500
},
{
"epoch": 35.19,
"learning_rate": 2.4103209756689523e-06,
"loss": 0.0156,
"step": 580000
},
{
"epoch": 35.22,
"learning_rate": 2.395151993204296e-06,
"loss": 0.0159,
"step": 580500
},
{
"epoch": 35.25,
"learning_rate": 2.380013348704569e-06,
"loss": 0.0163,
"step": 581000
},
{
"epoch": 35.28,
"learning_rate": 2.3648443662399128e-06,
"loss": 0.0162,
"step": 581500
},
{
"epoch": 35.31,
"learning_rate": 2.3496753837752565e-06,
"loss": 0.016,
"step": 582000
},
{
"epoch": 35.34,
"learning_rate": 2.3345064013106003e-06,
"loss": 0.0156,
"step": 582500
},
{
"epoch": 35.37,
"learning_rate": 2.3193677568108732e-06,
"loss": 0.0159,
"step": 583000
},
{
"epoch": 35.4,
"learning_rate": 2.304198774346217e-06,
"loss": 0.0158,
"step": 583500
},
{
"epoch": 35.43,
"learning_rate": 2.2890297918815607e-06,
"loss": 0.016,
"step": 584000
},
{
"epoch": 35.47,
"learning_rate": 2.2738608094169045e-06,
"loss": 0.0162,
"step": 584500
},
{
"epoch": 35.5,
"learning_rate": 2.2586918269522483e-06,
"loss": 0.0156,
"step": 585000
},
{
"epoch": 35.53,
"learning_rate": 2.243522844487592e-06,
"loss": 0.0158,
"step": 585500
},
{
"epoch": 35.56,
"learning_rate": 2.228384199987865e-06,
"loss": 0.0167,
"step": 586000
},
{
"epoch": 35.59,
"learning_rate": 2.2132152175232087e-06,
"loss": 0.0155,
"step": 586500
},
{
"epoch": 35.62,
"learning_rate": 2.1980462350585525e-06,
"loss": 0.0168,
"step": 587000
},
{
"epoch": 35.65,
"learning_rate": 2.1828772525938963e-06,
"loss": 0.0161,
"step": 587500
},
{
"epoch": 35.68,
"learning_rate": 2.16770827012924e-06,
"loss": 0.0156,
"step": 588000
},
{
"epoch": 35.71,
"learning_rate": 2.1525392876645838e-06,
"loss": 0.0162,
"step": 588500
},
{
"epoch": 35.74,
"learning_rate": 2.137370305199927e-06,
"loss": 0.0152,
"step": 589000
},
{
"epoch": 35.77,
"learning_rate": 2.1222013227352713e-06,
"loss": 0.0161,
"step": 589500
},
{
"epoch": 35.8,
"learning_rate": 2.1070323402706146e-06,
"loss": 0.0163,
"step": 590000
},
{
"epoch": 35.83,
"learning_rate": 2.091893695770888e-06,
"loss": 0.0161,
"step": 590500
},
{
"epoch": 35.86,
"learning_rate": 2.0767247133062313e-06,
"loss": 0.0158,
"step": 591000
},
{
"epoch": 35.89,
"learning_rate": 2.0615860688065047e-06,
"loss": 0.0164,
"step": 591500
},
{
"epoch": 35.92,
"learning_rate": 2.0464170863418485e-06,
"loss": 0.0159,
"step": 592000
},
{
"epoch": 35.95,
"learning_rate": 2.031248103877192e-06,
"loss": 0.0158,
"step": 592500
},
{
"epoch": 35.98,
"learning_rate": 2.016079121412536e-06,
"loss": 0.0162,
"step": 593000
},
{
"epoch": 36.0,
"eval_bleu": 87.7375,
"eval_gen_len": 14.7845,
"eval_loss": 0.22915604710578918,
"eval_runtime": 219.2242,
"eval_samples_per_second": 73.55,
"eval_steps_per_second": 2.299,
"step": 593316
},
{
"epoch": 36.01,
"learning_rate": 2.0009101389478797e-06,
"loss": 0.0157,
"step": 593500
},
{
"epoch": 36.04,
"learning_rate": 1.985741156483223e-06,
"loss": 0.0153,
"step": 594000
},
{
"epoch": 36.07,
"learning_rate": 1.9705721740185673e-06,
"loss": 0.0148,
"step": 594500
},
{
"epoch": 36.1,
"learning_rate": 1.9554031915539106e-06,
"loss": 0.0149,
"step": 595000
},
{
"epoch": 36.13,
"learning_rate": 1.9402645470541835e-06,
"loss": 0.0156,
"step": 595500
},
{
"epoch": 36.16,
"learning_rate": 1.9250955645895273e-06,
"loss": 0.0153,
"step": 596000
},
{
"epoch": 36.19,
"learning_rate": 1.909926582124871e-06,
"loss": 0.0156,
"step": 596500
},
{
"epoch": 36.22,
"learning_rate": 1.894757599660215e-06,
"loss": 0.0156,
"step": 597000
},
{
"epoch": 36.25,
"learning_rate": 1.8795886171955588e-06,
"loss": 0.0158,
"step": 597500
},
{
"epoch": 36.28,
"learning_rate": 1.8644196347309024e-06,
"loss": 0.0152,
"step": 598000
},
{
"epoch": 36.31,
"learning_rate": 1.8492506522662461e-06,
"loss": 0.0151,
"step": 598500
},
{
"epoch": 36.34,
"learning_rate": 1.8340816698015899e-06,
"loss": 0.0154,
"step": 599000
},
{
"epoch": 36.38,
"learning_rate": 1.8189126873369336e-06,
"loss": 0.0158,
"step": 599500
},
{
"epoch": 36.41,
"learning_rate": 1.8037740428372066e-06,
"loss": 0.0153,
"step": 600000
},
{
"epoch": 36.44,
"learning_rate": 1.7886050603725501e-06,
"loss": 0.0154,
"step": 600500
},
{
"epoch": 36.47,
"learning_rate": 1.773436077907894e-06,
"loss": 0.0148,
"step": 601000
},
{
"epoch": 36.5,
"learning_rate": 1.758297433408167e-06,
"loss": 0.0154,
"step": 601500
},
{
"epoch": 36.53,
"learning_rate": 1.74315878890844e-06,
"loss": 0.0156,
"step": 602000
},
{
"epoch": 36.56,
"learning_rate": 1.727989806443784e-06,
"loss": 0.0152,
"step": 602500
},
{
"epoch": 36.59,
"learning_rate": 1.7128511619440569e-06,
"loss": 0.015,
"step": 603000
},
{
"epoch": 36.62,
"learning_rate": 1.6976821794794006e-06,
"loss": 0.0149,
"step": 603500
},
{
"epoch": 36.65,
"learning_rate": 1.6825131970147444e-06,
"loss": 0.0152,
"step": 604000
},
{
"epoch": 36.68,
"learning_rate": 1.6673442145500882e-06,
"loss": 0.015,
"step": 604500
},
{
"epoch": 36.71,
"learning_rate": 1.6521752320854317e-06,
"loss": 0.015,
"step": 605000
},
{
"epoch": 36.74,
"learning_rate": 1.6370062496207757e-06,
"loss": 0.0157,
"step": 605500
},
{
"epoch": 36.77,
"learning_rate": 1.6218372671561192e-06,
"loss": 0.0158,
"step": 606000
},
{
"epoch": 36.8,
"learning_rate": 1.606668284691463e-06,
"loss": 0.0157,
"step": 606500
},
{
"epoch": 36.83,
"learning_rate": 1.5914993022268068e-06,
"loss": 0.0154,
"step": 607000
},
{
"epoch": 36.86,
"learning_rate": 1.5763303197621505e-06,
"loss": 0.0157,
"step": 607500
},
{
"epoch": 36.89,
"learning_rate": 1.5611916752624235e-06,
"loss": 0.0155,
"step": 608000
},
{
"epoch": 36.92,
"learning_rate": 1.5460226927977672e-06,
"loss": 0.0153,
"step": 608500
},
{
"epoch": 36.95,
"learning_rate": 1.5308537103331108e-06,
"loss": 0.016,
"step": 609000
},
{
"epoch": 36.98,
"learning_rate": 1.5156847278684548e-06,
"loss": 0.0157,
"step": 609500
},
{
"epoch": 37.0,
"eval_bleu": 87.8714,
"eval_gen_len": 14.8,
"eval_loss": 0.23112483322620392,
"eval_runtime": 219.854,
"eval_samples_per_second": 73.34,
"eval_steps_per_second": 2.292,
"step": 609797
},
{
"epoch": 37.01,
"learning_rate": 1.5005157454037985e-06,
"loss": 0.0144,
"step": 610000
},
{
"epoch": 37.04,
"learning_rate": 1.485346762939142e-06,
"loss": 0.0146,
"step": 610500
},
{
"epoch": 37.07,
"learning_rate": 1.470177780474486e-06,
"loss": 0.0145,
"step": 611000
},
{
"epoch": 37.1,
"learning_rate": 1.4550087980098296e-06,
"loss": 0.015,
"step": 611500
},
{
"epoch": 37.13,
"learning_rate": 1.4398701535101025e-06,
"loss": 0.0146,
"step": 612000
},
{
"epoch": 37.16,
"learning_rate": 1.4247011710454465e-06,
"loss": 0.0143,
"step": 612500
},
{
"epoch": 37.19,
"learning_rate": 1.40953218858079e-06,
"loss": 0.0152,
"step": 613000
},
{
"epoch": 37.22,
"learning_rate": 1.3943632061161338e-06,
"loss": 0.0153,
"step": 613500
},
{
"epoch": 37.26,
"learning_rate": 1.3791942236514778e-06,
"loss": 0.0146,
"step": 614000
},
{
"epoch": 37.29,
"learning_rate": 1.3640555791517507e-06,
"loss": 0.015,
"step": 614500
},
{
"epoch": 37.32,
"learning_rate": 1.3488865966870943e-06,
"loss": 0.0148,
"step": 615000
},
{
"epoch": 37.35,
"learning_rate": 1.3337479521873672e-06,
"loss": 0.0149,
"step": 615500
},
{
"epoch": 37.38,
"learning_rate": 1.3185789697227112e-06,
"loss": 0.0147,
"step": 616000
},
{
"epoch": 37.41,
"learning_rate": 1.303409987258055e-06,
"loss": 0.0153,
"step": 616500
},
{
"epoch": 37.44,
"learning_rate": 1.2882410047933985e-06,
"loss": 0.0139,
"step": 617000
},
{
"epoch": 37.47,
"learning_rate": 1.2730720223287425e-06,
"loss": 0.0161,
"step": 617500
},
{
"epoch": 37.5,
"learning_rate": 1.257903039864086e-06,
"loss": 0.0153,
"step": 618000
},
{
"epoch": 37.53,
"learning_rate": 1.2427340573994298e-06,
"loss": 0.0153,
"step": 618500
},
{
"epoch": 37.56,
"learning_rate": 1.2275954128997027e-06,
"loss": 0.0152,
"step": 619000
},
{
"epoch": 37.59,
"learning_rate": 1.2124264304350465e-06,
"loss": 0.015,
"step": 619500
},
{
"epoch": 37.62,
"learning_rate": 1.1972574479703902e-06,
"loss": 0.0147,
"step": 620000
},
{
"epoch": 37.65,
"learning_rate": 1.182088465505734e-06,
"loss": 0.0154,
"step": 620500
},
{
"epoch": 37.68,
"learning_rate": 1.1669194830410778e-06,
"loss": 0.0144,
"step": 621000
},
{
"epoch": 37.71,
"learning_rate": 1.1517505005764213e-06,
"loss": 0.015,
"step": 621500
},
{
"epoch": 37.74,
"learning_rate": 1.136581518111765e-06,
"loss": 0.0155,
"step": 622000
},
{
"epoch": 37.77,
"learning_rate": 1.121412535647109e-06,
"loss": 0.0147,
"step": 622500
},
{
"epoch": 37.8,
"learning_rate": 1.1062435531824526e-06,
"loss": 0.0148,
"step": 623000
},
{
"epoch": 37.83,
"learning_rate": 1.0910745707177964e-06,
"loss": 0.0149,
"step": 623500
},
{
"epoch": 37.86,
"learning_rate": 1.0759055882531401e-06,
"loss": 0.0148,
"step": 624000
},
{
"epoch": 37.89,
"learning_rate": 1.0607366057884839e-06,
"loss": 0.0151,
"step": 624500
},
{
"epoch": 37.92,
"learning_rate": 1.0455979612887568e-06,
"loss": 0.0152,
"step": 625000
},
{
"epoch": 37.95,
"learning_rate": 1.0304289788241006e-06,
"loss": 0.0146,
"step": 625500
},
{
"epoch": 37.98,
"learning_rate": 1.0152599963594443e-06,
"loss": 0.0153,
"step": 626000
},
{
"epoch": 38.0,
"eval_bleu": 87.8712,
"eval_gen_len": 14.799,
"eval_loss": 0.23195865750312805,
"eval_runtime": 218.5324,
"eval_samples_per_second": 73.783,
"eval_steps_per_second": 2.306,
"step": 626278
},
{
"epoch": 38.01,
"learning_rate": 1.0000910138947881e-06,
"loss": 0.0147,
"step": 626500
},
{
"epoch": 38.04,
"learning_rate": 9.849220314301319e-07,
"loss": 0.0142,
"step": 627000
},
{
"epoch": 38.07,
"learning_rate": 9.697530489654754e-07,
"loss": 0.0151,
"step": 627500
},
{
"epoch": 38.1,
"learning_rate": 9.545840665008192e-07,
"loss": 0.0151,
"step": 628000
},
{
"epoch": 38.13,
"learning_rate": 9.394454220010922e-07,
"loss": 0.0143,
"step": 628500
},
{
"epoch": 38.17,
"learning_rate": 9.242764395364359e-07,
"loss": 0.0147,
"step": 629000
},
{
"epoch": 38.2,
"learning_rate": 9.091074570717796e-07,
"loss": 0.0144,
"step": 629500
},
{
"epoch": 38.23,
"learning_rate": 8.939384746071235e-07,
"loss": 0.0148,
"step": 630000
},
{
"epoch": 38.26,
"learning_rate": 8.787998301073964e-07,
"loss": 0.015,
"step": 630500
},
{
"epoch": 38.29,
"learning_rate": 8.636308476427402e-07,
"loss": 0.0138,
"step": 631000
},
{
"epoch": 38.32,
"learning_rate": 8.484618651780839e-07,
"loss": 0.0146,
"step": 631500
},
{
"epoch": 38.35,
"learning_rate": 8.332928827134276e-07,
"loss": 0.0144,
"step": 632000
},
{
"epoch": 38.38,
"learning_rate": 8.181239002487714e-07,
"loss": 0.0144,
"step": 632500
},
{
"epoch": 38.41,
"learning_rate": 8.02954917784115e-07,
"loss": 0.0142,
"step": 633000
},
{
"epoch": 38.44,
"learning_rate": 7.878466112493175e-07,
"loss": 0.0145,
"step": 633500
},
{
"epoch": 38.47,
"learning_rate": 7.726776287846611e-07,
"loss": 0.015,
"step": 634000
},
{
"epoch": 38.5,
"learning_rate": 7.575086463200049e-07,
"loss": 0.0147,
"step": 634500
},
{
"epoch": 38.53,
"learning_rate": 7.423396638553485e-07,
"loss": 0.0153,
"step": 635000
},
{
"epoch": 38.56,
"learning_rate": 7.271706813906924e-07,
"loss": 0.0151,
"step": 635500
},
{
"epoch": 38.59,
"learning_rate": 7.120016989260362e-07,
"loss": 0.0143,
"step": 636000
},
{
"epoch": 38.62,
"learning_rate": 6.968327164613798e-07,
"loss": 0.0143,
"step": 636500
},
{
"epoch": 38.65,
"learning_rate": 6.816637339967236e-07,
"loss": 0.015,
"step": 637000
},
{
"epoch": 38.68,
"learning_rate": 6.664947515320672e-07,
"loss": 0.0145,
"step": 637500
},
{
"epoch": 38.71,
"learning_rate": 6.51325769067411e-07,
"loss": 0.0142,
"step": 638000
},
{
"epoch": 38.74,
"learning_rate": 6.361871245676841e-07,
"loss": 0.0149,
"step": 638500
},
{
"epoch": 38.77,
"learning_rate": 6.210181421030278e-07,
"loss": 0.0146,
"step": 639000
},
{
"epoch": 38.8,
"learning_rate": 6.058491596383715e-07,
"loss": 0.015,
"step": 639500
},
{
"epoch": 38.83,
"learning_rate": 5.906801771737152e-07,
"loss": 0.0147,
"step": 640000
},
{
"epoch": 38.86,
"learning_rate": 5.755415326739883e-07,
"loss": 0.0146,
"step": 640500
},
{
"epoch": 38.89,
"learning_rate": 5.603725502093319e-07,
"loss": 0.0147,
"step": 641000
},
{
"epoch": 38.92,
"learning_rate": 5.452035677446757e-07,
"loss": 0.0147,
"step": 641500
},
{
"epoch": 38.95,
"learning_rate": 5.300345852800195e-07,
"loss": 0.0149,
"step": 642000
},
{
"epoch": 38.98,
"learning_rate": 5.148656028153632e-07,
"loss": 0.0149,
"step": 642500
},
{
"epoch": 39.0,
"eval_bleu": 87.8655,
"eval_gen_len": 14.7899,
"eval_loss": 0.2320941537618637,
"eval_runtime": 218.6389,
"eval_samples_per_second": 73.747,
"eval_steps_per_second": 2.305,
"step": 642759
},
{
"epoch": 39.01,
"learning_rate": 4.997269583156363e-07,
"loss": 0.0142,
"step": 643000
},
{
"epoch": 39.04,
"learning_rate": 4.845579758509799e-07,
"loss": 0.0143,
"step": 643500
},
{
"epoch": 39.08,
"learning_rate": 4.693889933863237e-07,
"loss": 0.0145,
"step": 644000
},
{
"epoch": 39.11,
"learning_rate": 4.542200109216674e-07,
"loss": 0.0145,
"step": 644500
},
{
"epoch": 39.14,
"learning_rate": 4.390510284570111e-07,
"loss": 0.0142,
"step": 645000
},
{
"epoch": 39.17,
"learning_rate": 4.238820459923549e-07,
"loss": 0.0145,
"step": 645500
},
{
"epoch": 39.2,
"learning_rate": 4.087130635276986e-07,
"loss": 0.0141,
"step": 646000
},
{
"epoch": 39.23,
"learning_rate": 3.9354408106304233e-07,
"loss": 0.0143,
"step": 646500
},
{
"epoch": 39.26,
"learning_rate": 3.7840543656331537e-07,
"loss": 0.0143,
"step": 647000
},
{
"epoch": 39.29,
"learning_rate": 3.632364540986591e-07,
"loss": 0.0149,
"step": 647500
},
{
"epoch": 39.32,
"learning_rate": 3.480674716340028e-07,
"loss": 0.0146,
"step": 648000
},
{
"epoch": 39.35,
"learning_rate": 3.3289848916934655e-07,
"loss": 0.0142,
"step": 648500
},
{
"epoch": 39.38,
"learning_rate": 3.1772950670469026e-07,
"loss": 0.0145,
"step": 649000
},
{
"epoch": 39.41,
"learning_rate": 3.0256052424003397e-07,
"loss": 0.0145,
"step": 649500
},
{
"epoch": 39.44,
"learning_rate": 2.8739154177537773e-07,
"loss": 0.0141,
"step": 650000
},
{
"epoch": 39.47,
"learning_rate": 2.7225289727565077e-07,
"loss": 0.0146,
"step": 650500
},
{
"epoch": 39.5,
"learning_rate": 2.5708391481099453e-07,
"loss": 0.0146,
"step": 651000
},
{
"epoch": 39.53,
"learning_rate": 2.4191493234633824e-07,
"loss": 0.0141,
"step": 651500
},
{
"epoch": 39.56,
"learning_rate": 2.2674594988168195e-07,
"loss": 0.0146,
"step": 652000
},
{
"epoch": 39.59,
"learning_rate": 2.115769674170257e-07,
"loss": 0.0147,
"step": 652500
},
{
"epoch": 39.62,
"learning_rate": 1.964383229172987e-07,
"loss": 0.014,
"step": 653000
},
{
"epoch": 39.65,
"learning_rate": 1.8129967841757177e-07,
"loss": 0.0138,
"step": 653500
},
{
"epoch": 39.68,
"learning_rate": 1.6613069595291548e-07,
"loss": 0.0148,
"step": 654000
},
{
"epoch": 39.71,
"learning_rate": 1.5096171348825922e-07,
"loss": 0.0145,
"step": 654500
},
{
"epoch": 39.74,
"learning_rate": 1.3579273102360295e-07,
"loss": 0.0142,
"step": 655000
},
{
"epoch": 39.77,
"learning_rate": 1.2062374855894669e-07,
"loss": 0.0146,
"step": 655500
},
{
"epoch": 39.8,
"learning_rate": 1.0545476609429041e-07,
"loss": 0.0144,
"step": 656000
},
{
"epoch": 39.83,
"learning_rate": 9.028578362963412e-08,
"loss": 0.014,
"step": 656500
},
{
"epoch": 39.86,
"learning_rate": 7.514713912990717e-08,
"loss": 0.0141,
"step": 657000
},
{
"epoch": 39.89,
"learning_rate": 5.99781566652509e-08,
"loss": 0.0139,
"step": 657500
},
{
"epoch": 39.92,
"learning_rate": 4.4809174200594624e-08,
"loss": 0.0147,
"step": 658000
},
{
"epoch": 39.96,
"learning_rate": 2.9640191735938357e-08,
"loss": 0.0144,
"step": 658500
},
{
"epoch": 39.99,
"learning_rate": 1.4471209271282084e-08,
"loss": 0.0144,
"step": 659000
},
{
"epoch": 40.0,
"eval_bleu": 87.8962,
"eval_gen_len": 14.792,
"eval_loss": 0.23238389194011688,
"eval_runtime": 218.3118,
"eval_samples_per_second": 73.858,
"eval_steps_per_second": 2.309,
"step": 659240
}
],
"max_steps": 659240,
"num_train_epochs": 40,
"total_flos": 1.3263324878544568e+18,
"trial_name": null,
"trial_params": null
}