joelniklaus's picture
Training in progress, step 200000
b8aef66
raw
history blame
25.6 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 95.0063,
"global_step": 200000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1e-05,
"loss": 6.365,
"step": 1000
},
{
"epoch": 0.01,
"learning_rate": 2e-05,
"loss": 4.54,
"step": 2000
},
{
"epoch": 1.0,
"learning_rate": 3e-05,
"loss": 3.9405,
"step": 3000
},
{
"epoch": 1.01,
"learning_rate": 4e-05,
"loss": 3.6444,
"step": 4000
},
{
"epoch": 2.0,
"learning_rate": 5e-05,
"loss": 3.1408,
"step": 5000
},
{
"epoch": 2.01,
"learning_rate": 6e-05,
"loss": 2.3019,
"step": 6000
},
{
"epoch": 3.0,
"learning_rate": 7e-05,
"loss": 1.802,
"step": 7000
},
{
"epoch": 3.01,
"learning_rate": 8e-05,
"loss": 1.6724,
"step": 8000
},
{
"epoch": 4.0,
"learning_rate": 9e-05,
"loss": 1.5194,
"step": 9000
},
{
"epoch": 4.01,
"learning_rate": 0.0001,
"loss": 1.4654,
"step": 10000
},
{
"epoch": 5.0,
"learning_rate": 9.999316524962345e-05,
"loss": 1.4209,
"step": 11000
},
{
"epoch": 5.01,
"learning_rate": 9.997266286704631e-05,
"loss": 1.3222,
"step": 12000
},
{
"epoch": 6.0,
"learning_rate": 9.993849845741524e-05,
"loss": 1.3633,
"step": 13000
},
{
"epoch": 6.01,
"learning_rate": 9.989068136093873e-05,
"loss": 1.2351,
"step": 14000
},
{
"epoch": 7.0,
"learning_rate": 9.98292246503335e-05,
"loss": 1.3078,
"step": 15000
},
{
"epoch": 7.01,
"learning_rate": 9.975414512725057e-05,
"loss": 1.1871,
"step": 16000
},
{
"epoch": 8.0,
"learning_rate": 9.966546331768191e-05,
"loss": 1.2577,
"step": 17000
},
{
"epoch": 8.01,
"learning_rate": 9.956320346634876e-05,
"loss": 1.1666,
"step": 18000
},
{
"epoch": 9.0,
"learning_rate": 9.944739353007344e-05,
"loss": 1.2102,
"step": 19000
},
{
"epoch": 9.01,
"learning_rate": 9.931806517013612e-05,
"loss": 1.1394,
"step": 20000
},
{
"epoch": 10.0,
"learning_rate": 9.917525374361912e-05,
"loss": 1.1911,
"step": 21000
},
{
"epoch": 10.01,
"learning_rate": 9.901899829374047e-05,
"loss": 1.1113,
"step": 22000
},
{
"epoch": 10.01,
"learning_rate": 9.884934153917997e-05,
"loss": 1.1768,
"step": 23000
},
{
"epoch": 11.0,
"learning_rate": 9.86663298624003e-05,
"loss": 1.0943,
"step": 24000
},
{
"epoch": 11.01,
"learning_rate": 9.847001329696653e-05,
"loss": 1.1477,
"step": 25000
},
{
"epoch": 12.0,
"learning_rate": 9.826044551386744e-05,
"loss": 1.0749,
"step": 26000
},
{
"epoch": 12.01,
"learning_rate": 9.803768380684242e-05,
"loss": 1.1274,
"step": 27000
},
{
"epoch": 13.0,
"learning_rate": 9.780178907671789e-05,
"loss": 1.0561,
"step": 28000
},
{
"epoch": 13.01,
"learning_rate": 9.755282581475769e-05,
"loss": 1.1045,
"step": 29000
},
{
"epoch": 14.0,
"learning_rate": 9.729086208503174e-05,
"loss": 1.0489,
"step": 30000
},
{
"epoch": 14.01,
"learning_rate": 9.701596950580806e-05,
"loss": 1.0863,
"step": 31000
},
{
"epoch": 15.0,
"learning_rate": 9.672822322997305e-05,
"loss": 1.0483,
"step": 32000
},
{
"epoch": 15.01,
"learning_rate": 9.642770192448536e-05,
"loss": 1.0622,
"step": 33000
},
{
"epoch": 16.0,
"learning_rate": 9.611448774886924e-05,
"loss": 1.0711,
"step": 34000
},
{
"epoch": 16.01,
"learning_rate": 9.578866633275288e-05,
"loss": 1.0189,
"step": 35000
},
{
"epoch": 17.0,
"learning_rate": 9.545032675245813e-05,
"loss": 1.0806,
"step": 36000
},
{
"epoch": 17.01,
"learning_rate": 9.509956150664796e-05,
"loss": 0.9976,
"step": 37000
},
{
"epoch": 18.0,
"learning_rate": 9.473646649103818e-05,
"loss": 1.0707,
"step": 38000
},
{
"epoch": 18.01,
"learning_rate": 9.43611409721806e-05,
"loss": 0.9852,
"step": 39000
},
{
"epoch": 19.0,
"learning_rate": 9.397368756032445e-05,
"loss": 1.057,
"step": 40000
},
{
"epoch": 19.01,
"learning_rate": 9.357421218136386e-05,
"loss": 0.988,
"step": 41000
},
{
"epoch": 20.0,
"learning_rate": 9.316282404787871e-05,
"loss": 1.0394,
"step": 42000
},
{
"epoch": 20.01,
"learning_rate": 9.273963562927695e-05,
"loss": 0.9825,
"step": 43000
},
{
"epoch": 21.0,
"learning_rate": 9.230476262104677e-05,
"loss": 1.0418,
"step": 44000
},
{
"epoch": 21.01,
"learning_rate": 9.185832391312644e-05,
"loss": 0.9664,
"step": 45000
},
{
"epoch": 21.01,
"learning_rate": 9.140044155740101e-05,
"loss": 1.0387,
"step": 46000
},
{
"epoch": 22.0,
"learning_rate": 9.093124073433463e-05,
"loss": 0.9689,
"step": 47000
},
{
"epoch": 22.01,
"learning_rate": 9.045084971874738e-05,
"loss": 1.0214,
"step": 48000
},
{
"epoch": 23.0,
"learning_rate": 8.995939984474624e-05,
"loss": 0.9581,
"step": 49000
},
{
"epoch": 23.01,
"learning_rate": 8.945702546981969e-05,
"loss": 1.013,
"step": 50000
},
{
"epoch": 23.01,
"eval_loss": 0.7245611548423767,
"eval_runtime": 16.7835,
"eval_samples_per_second": 297.912,
"eval_steps_per_second": 2.383,
"step": 50000
},
{
"epoch": 24.0,
"learning_rate": 8.894386393810563e-05,
"loss": 0.9485,
"step": 51000
},
{
"epoch": 24.01,
"learning_rate": 8.842005554284296e-05,
"loss": 0.9995,
"step": 52000
},
{
"epoch": 25.0,
"learning_rate": 8.788574348801675e-05,
"loss": 0.9509,
"step": 53000
},
{
"epoch": 25.01,
"learning_rate": 8.73410738492077e-05,
"loss": 0.9897,
"step": 54000
},
{
"epoch": 26.0,
"learning_rate": 8.678619553365659e-05,
"loss": 0.9599,
"step": 55000
},
{
"epoch": 26.01,
"learning_rate": 8.622126023955446e-05,
"loss": 0.968,
"step": 56000
},
{
"epoch": 27.0,
"learning_rate": 8.564642241456986e-05,
"loss": 0.9889,
"step": 57000
},
{
"epoch": 27.01,
"learning_rate": 8.506183921362443e-05,
"loss": 0.9331,
"step": 58000
},
{
"epoch": 28.0,
"learning_rate": 8.44676704559283e-05,
"loss": 0.9972,
"step": 59000
},
{
"epoch": 28.01,
"learning_rate": 8.386407858128706e-05,
"loss": 0.9208,
"step": 60000
},
{
"epoch": 29.0,
"learning_rate": 8.32512286056924e-05,
"loss": 0.9923,
"step": 61000
},
{
"epoch": 29.01,
"learning_rate": 8.262928807620843e-05,
"loss": 0.9152,
"step": 62000
},
{
"epoch": 30.0,
"learning_rate": 8.199842702516583e-05,
"loss": 0.981,
"step": 63000
},
{
"epoch": 30.01,
"learning_rate": 8.135881792367686e-05,
"loss": 0.9202,
"step": 64000
},
{
"epoch": 31.0,
"learning_rate": 8.07106356344834e-05,
"loss": 0.9697,
"step": 65000
},
{
"epoch": 31.01,
"learning_rate": 8.005405736415126e-05,
"loss": 0.9151,
"step": 66000
},
{
"epoch": 32.0,
"learning_rate": 7.938926261462366e-05,
"loss": 0.9762,
"step": 67000
},
{
"epoch": 32.01,
"learning_rate": 7.871643313414718e-05,
"loss": 0.9036,
"step": 68000
},
{
"epoch": 32.01,
"learning_rate": 7.803575286758364e-05,
"loss": 0.9742,
"step": 69000
},
{
"epoch": 33.0,
"learning_rate": 7.734740790612136e-05,
"loss": 0.9105,
"step": 70000
},
{
"epoch": 33.01,
"learning_rate": 7.66515864363997e-05,
"loss": 0.9611,
"step": 71000
},
{
"epoch": 34.0,
"learning_rate": 7.594847868906076e-05,
"loss": 0.8994,
"step": 72000
},
{
"epoch": 34.01,
"learning_rate": 7.52382768867422e-05,
"loss": 0.9551,
"step": 73000
},
{
"epoch": 35.0,
"learning_rate": 7.452117519152542e-05,
"loss": 0.8943,
"step": 74000
},
{
"epoch": 35.01,
"learning_rate": 7.379736965185368e-05,
"loss": 0.9445,
"step": 75000
},
{
"epoch": 36.0,
"learning_rate": 7.30670581489344e-05,
"loss": 0.896,
"step": 76000
},
{
"epoch": 36.01,
"learning_rate": 7.233044034264034e-05,
"loss": 0.9371,
"step": 77000
},
{
"epoch": 37.0,
"learning_rate": 7.158771761692464e-05,
"loss": 0.911,
"step": 78000
},
{
"epoch": 37.01,
"learning_rate": 7.083909302476453e-05,
"loss": 0.9143,
"step": 79000
},
{
"epoch": 38.0,
"learning_rate": 7.008477123264848e-05,
"loss": 0.9433,
"step": 80000
},
{
"epoch": 38.01,
"learning_rate": 6.932495846462261e-05,
"loss": 0.881,
"step": 81000
},
{
"epoch": 39.0,
"learning_rate": 6.855986244591104e-05,
"loss": 0.9461,
"step": 82000
},
{
"epoch": 39.01,
"learning_rate": 6.778969234612584e-05,
"loss": 0.8759,
"step": 83000
},
{
"epoch": 40.0,
"learning_rate": 6.701465872208216e-05,
"loss": 0.944,
"step": 84000
},
{
"epoch": 40.01,
"learning_rate": 6.623497346023418e-05,
"loss": 0.8766,
"step": 85000
},
{
"epoch": 41.0,
"learning_rate": 6.545084971874738e-05,
"loss": 0.9291,
"step": 86000
},
{
"epoch": 41.01,
"learning_rate": 6.466250186922325e-05,
"loss": 0.8758,
"step": 87000
},
{
"epoch": 42.0,
"learning_rate": 6.387014543809223e-05,
"loss": 0.9281,
"step": 88000
},
{
"epoch": 42.01,
"learning_rate": 6.307399704769099e-05,
"loss": 0.8725,
"step": 89000
},
{
"epoch": 43.0,
"learning_rate": 6.227427435703997e-05,
"loss": 0.9344,
"step": 90000
},
{
"epoch": 43.01,
"learning_rate": 6.147119600233758e-05,
"loss": 0.8645,
"step": 91000
},
{
"epoch": 43.01,
"learning_rate": 6.066498153718735e-05,
"loss": 0.9319,
"step": 92000
},
{
"epoch": 44.0,
"learning_rate": 5.985585137257401e-05,
"loss": 0.8698,
"step": 93000
},
{
"epoch": 44.01,
"learning_rate": 5.90440267166055e-05,
"loss": 0.9219,
"step": 94000
},
{
"epoch": 45.0,
"learning_rate": 5.8229729514036705e-05,
"loss": 0.8574,
"step": 95000
},
{
"epoch": 45.01,
"learning_rate": 5.74131823855921e-05,
"loss": 0.9154,
"step": 96000
},
{
"epoch": 46.0,
"learning_rate": 5.6594608567103456e-05,
"loss": 0.8572,
"step": 97000
},
{
"epoch": 46.01,
"learning_rate": 5.577423184847932e-05,
"loss": 0.9074,
"step": 98000
},
{
"epoch": 47.0,
"learning_rate": 5.495227651252315e-05,
"loss": 0.8593,
"step": 99000
},
{
"epoch": 47.01,
"learning_rate": 5.4128967273616625e-05,
"loss": 0.8993,
"step": 100000
},
{
"epoch": 47.01,
"eval_loss": 0.6356155276298523,
"eval_runtime": 10.4482,
"eval_samples_per_second": 478.553,
"eval_steps_per_second": 3.828,
"step": 100000
},
{
"epoch": 48.0,
"learning_rate": 5.330452921628497e-05,
"loss": 0.8789,
"step": 101000
},
{
"epoch": 48.01,
"learning_rate": 5.247918773366112e-05,
"loss": 0.8769,
"step": 102000
},
{
"epoch": 49.0,
"learning_rate": 5.165316846586541e-05,
"loss": 0.9084,
"step": 103000
},
{
"epoch": 49.01,
"learning_rate": 5.0826697238317935e-05,
"loss": 0.8454,
"step": 104000
},
{
"epoch": 50.0,
"learning_rate": 5e-05,
"loss": 0.9128,
"step": 105000
},
{
"epoch": 50.01,
"learning_rate": 4.917330276168208e-05,
"loss": 0.8434,
"step": 106000
},
{
"epoch": 51.0,
"learning_rate": 4.834683153413459e-05,
"loss": 0.9092,
"step": 107000
},
{
"epoch": 51.01,
"learning_rate": 4.7520812266338885e-05,
"loss": 0.8447,
"step": 108000
},
{
"epoch": 52.0,
"learning_rate": 4.669547078371504e-05,
"loss": 0.8957,
"step": 109000
},
{
"epoch": 52.01,
"learning_rate": 4.5871032726383386e-05,
"loss": 0.8444,
"step": 110000
},
{
"epoch": 53.0,
"learning_rate": 4.504772348747687e-05,
"loss": 0.8977,
"step": 111000
},
{
"epoch": 53.01,
"learning_rate": 4.4225768151520694e-05,
"loss": 0.8415,
"step": 112000
},
{
"epoch": 54.0,
"learning_rate": 4.3405391432896555e-05,
"loss": 0.9041,
"step": 113000
},
{
"epoch": 54.01,
"learning_rate": 4.2586817614407895e-05,
"loss": 0.8351,
"step": 114000
},
{
"epoch": 54.01,
"learning_rate": 4.17702704859633e-05,
"loss": 0.9016,
"step": 115000
},
{
"epoch": 55.0,
"learning_rate": 4.095597328339452e-05,
"loss": 0.8389,
"step": 116000
},
{
"epoch": 55.01,
"learning_rate": 4.0144148627425993e-05,
"loss": 0.8926,
"step": 117000
},
{
"epoch": 56.0,
"learning_rate": 3.933501846281267e-05,
"loss": 0.8277,
"step": 118000
},
{
"epoch": 56.01,
"learning_rate": 3.852880399766243e-05,
"loss": 0.8866,
"step": 119000
},
{
"epoch": 57.0,
"learning_rate": 3.772572564296005e-05,
"loss": 0.83,
"step": 120000
},
{
"epoch": 57.01,
"learning_rate": 3.6926002952309016e-05,
"loss": 0.8796,
"step": 121000
},
{
"epoch": 58.0,
"learning_rate": 3.612985456190778e-05,
"loss": 0.8335,
"step": 122000
},
{
"epoch": 58.01,
"learning_rate": 3.533749813077677e-05,
"loss": 0.8703,
"step": 123000
},
{
"epoch": 59.0,
"learning_rate": 3.4549150281252636e-05,
"loss": 0.8569,
"step": 124000
},
{
"epoch": 59.01,
"learning_rate": 3.3765026539765834e-05,
"loss": 0.8457,
"step": 125000
},
{
"epoch": 60.0,
"learning_rate": 3.298534127791785e-05,
"loss": 0.8837,
"step": 126000
},
{
"epoch": 60.01,
"learning_rate": 3.221030765387417e-05,
"loss": 0.8192,
"step": 127000
},
{
"epoch": 61.0,
"learning_rate": 3.144013755408895e-05,
"loss": 0.8873,
"step": 128000
},
{
"epoch": 61.01,
"learning_rate": 3.0675041535377405e-05,
"loss": 0.8156,
"step": 129000
},
{
"epoch": 62.0,
"learning_rate": 2.991522876735154e-05,
"loss": 0.8843,
"step": 130000
},
{
"epoch": 62.01,
"learning_rate": 2.916090697523549e-05,
"loss": 0.8205,
"step": 131000
},
{
"epoch": 63.0,
"learning_rate": 2.8412282383075363e-05,
"loss": 0.8709,
"step": 132000
},
{
"epoch": 63.01,
"learning_rate": 2.766955965735968e-05,
"loss": 0.8216,
"step": 133000
},
{
"epoch": 64.0,
"learning_rate": 2.693294185106562e-05,
"loss": 0.8749,
"step": 134000
},
{
"epoch": 64.01,
"learning_rate": 2.6202630348146324e-05,
"loss": 0.8169,
"step": 135000
},
{
"epoch": 65.0,
"learning_rate": 2.547882480847461e-05,
"loss": 0.883,
"step": 136000
},
{
"epoch": 65.01,
"learning_rate": 2.476172311325783e-05,
"loss": 0.8111,
"step": 137000
},
{
"epoch": 65.01,
"learning_rate": 2.405152131093926e-05,
"loss": 0.8786,
"step": 138000
},
{
"epoch": 66.0,
"learning_rate": 2.3348413563600325e-05,
"loss": 0.8158,
"step": 139000
},
{
"epoch": 66.01,
"learning_rate": 2.2652592093878666e-05,
"loss": 0.874,
"step": 140000
},
{
"epoch": 67.0,
"learning_rate": 2.196424713241637e-05,
"loss": 0.8035,
"step": 141000
},
{
"epoch": 67.01,
"learning_rate": 2.128356686585282e-05,
"loss": 0.8646,
"step": 142000
},
{
"epoch": 68.0,
"learning_rate": 2.061073738537635e-05,
"loss": 0.8097,
"step": 143000
},
{
"epoch": 68.01,
"learning_rate": 1.9945942635848748e-05,
"loss": 0.8582,
"step": 144000
},
{
"epoch": 69.0,
"learning_rate": 1.928936436551661e-05,
"loss": 0.8153,
"step": 145000
},
{
"epoch": 69.01,
"learning_rate": 1.8641182076323148e-05,
"loss": 0.8499,
"step": 146000
},
{
"epoch": 70.0,
"learning_rate": 1.800157297483417e-05,
"loss": 0.8435,
"step": 147000
},
{
"epoch": 70.01,
"learning_rate": 1.7370711923791567e-05,
"loss": 0.8214,
"step": 148000
},
{
"epoch": 71.0,
"learning_rate": 1.6748771394307585e-05,
"loss": 0.8673,
"step": 149000
},
{
"epoch": 71.01,
"learning_rate": 1.6135921418712956e-05,
"loss": 0.8031,
"step": 150000
},
{
"epoch": 71.01,
"eval_loss": 0.599342405796051,
"eval_runtime": 10.8341,
"eval_samples_per_second": 461.505,
"eval_steps_per_second": 3.692,
"step": 150000
},
{
"epoch": 72.0,
"learning_rate": 1.553232954407171e-05,
"loss": 0.8685,
"step": 151000
},
{
"epoch": 72.01,
"learning_rate": 1.4938160786375572e-05,
"loss": 0.7981,
"step": 152000
},
{
"epoch": 73.0,
"learning_rate": 1.435357758543015e-05,
"loss": 0.8652,
"step": 153000
},
{
"epoch": 73.01,
"learning_rate": 1.3778739760445552e-05,
"loss": 0.8056,
"step": 154000
},
{
"epoch": 74.0,
"learning_rate": 1.3213804466343421e-05,
"loss": 0.8543,
"step": 155000
},
{
"epoch": 74.01,
"learning_rate": 1.2658926150792322e-05,
"loss": 0.8036,
"step": 156000
},
{
"epoch": 75.0,
"learning_rate": 1.2114256511983274e-05,
"loss": 0.8601,
"step": 157000
},
{
"epoch": 75.01,
"learning_rate": 1.157994445715706e-05,
"loss": 0.8004,
"step": 158000
},
{
"epoch": 76.0,
"learning_rate": 1.1056136061894384e-05,
"loss": 0.8691,
"step": 159000
},
{
"epoch": 76.01,
"learning_rate": 1.0542974530180327e-05,
"loss": 0.7977,
"step": 160000
},
{
"epoch": 76.01,
"learning_rate": 1.0040600155253765e-05,
"loss": 0.8628,
"step": 161000
},
{
"epoch": 77.0,
"learning_rate": 9.549150281252633e-06,
"loss": 0.7984,
"step": 162000
},
{
"epoch": 77.01,
"learning_rate": 9.068759265665384e-06,
"loss": 0.8598,
"step": 163000
},
{
"epoch": 78.0,
"learning_rate": 8.599558442598998e-06,
"loss": 0.7912,
"step": 164000
},
{
"epoch": 78.01,
"learning_rate": 8.141676086873572e-06,
"loss": 0.851,
"step": 165000
},
{
"epoch": 79.0,
"learning_rate": 7.695237378953223e-06,
"loss": 0.7979,
"step": 166000
},
{
"epoch": 79.01,
"learning_rate": 7.260364370723044e-06,
"loss": 0.8447,
"step": 167000
},
{
"epoch": 80.0,
"learning_rate": 6.837175952121306e-06,
"loss": 0.809,
"step": 168000
},
{
"epoch": 80.01,
"learning_rate": 6.425787818636131e-06,
"loss": 0.8335,
"step": 169000
},
{
"epoch": 81.0,
"learning_rate": 6.026312439675552e-06,
"loss": 0.834,
"step": 170000
},
{
"epoch": 81.01,
"learning_rate": 5.6388590278194096e-06,
"loss": 0.8079,
"step": 171000
},
{
"epoch": 82.0,
"learning_rate": 5.263533508961827e-06,
"loss": 0.8555,
"step": 172000
},
{
"epoch": 82.01,
"learning_rate": 4.900438493352055e-06,
"loss": 0.7925,
"step": 173000
},
{
"epoch": 83.0,
"learning_rate": 4.549673247541875e-06,
"loss": 0.8589,
"step": 174000
},
{
"epoch": 83.01,
"learning_rate": 4.2113336672471245e-06,
"loss": 0.7873,
"step": 175000
},
{
"epoch": 84.0,
"learning_rate": 3.885512251130763e-06,
"loss": 0.8574,
"step": 176000
},
{
"epoch": 84.01,
"learning_rate": 3.5722980755146517e-06,
"loss": 0.7967,
"step": 177000
},
{
"epoch": 85.0,
"learning_rate": 3.271776770026963e-06,
"loss": 0.846,
"step": 178000
},
{
"epoch": 85.01,
"learning_rate": 2.9840304941919415e-06,
"loss": 0.796,
"step": 179000
},
{
"epoch": 86.0,
"learning_rate": 2.7091379149682685e-06,
"loss": 0.8528,
"step": 180000
},
{
"epoch": 86.01,
"learning_rate": 2.4471741852423237e-06,
"loss": 0.7928,
"step": 181000
},
{
"epoch": 86.01,
"learning_rate": 2.1982109232821178e-06,
"loss": 0.862,
"step": 182000
},
{
"epoch": 87.0,
"learning_rate": 1.962316193157593e-06,
"loss": 0.7927,
"step": 183000
},
{
"epoch": 87.01,
"learning_rate": 1.7395544861325718e-06,
"loss": 0.855,
"step": 184000
},
{
"epoch": 88.0,
"learning_rate": 1.5299867030334814e-06,
"loss": 0.7916,
"step": 185000
},
{
"epoch": 88.01,
"learning_rate": 1.333670137599713e-06,
"loss": 0.8526,
"step": 186000
},
{
"epoch": 89.0,
"learning_rate": 1.1506584608200367e-06,
"loss": 0.7861,
"step": 187000
},
{
"epoch": 89.01,
"learning_rate": 9.810017062595322e-07,
"loss": 0.8453,
"step": 188000
},
{
"epoch": 90.0,
"learning_rate": 8.247462563808817e-07,
"loss": 0.7931,
"step": 189000
},
{
"epoch": 90.01,
"learning_rate": 6.819348298638839e-07,
"loss": 0.8392,
"step": 190000
},
{
"epoch": 91.0,
"learning_rate": 5.526064699265753e-07,
"loss": 0.8067,
"step": 191000
},
{
"epoch": 91.01,
"learning_rate": 4.367965336512403e-07,
"loss": 0.8269,
"step": 192000
},
{
"epoch": 92.0,
"learning_rate": 3.3453668231809286e-07,
"loss": 0.8336,
"step": 193000
},
{
"epoch": 92.01,
"learning_rate": 2.458548727494292e-07,
"loss": 0.8028,
"step": 194000
},
{
"epoch": 93.0,
"learning_rate": 1.7077534966650766e-07,
"loss": 0.8539,
"step": 195000
},
{
"epoch": 93.01,
"learning_rate": 1.0931863906127327e-07,
"loss": 0.7885,
"step": 196000
},
{
"epoch": 94.0,
"learning_rate": 6.150154258476315e-08,
"loss": 0.8556,
"step": 197000
},
{
"epoch": 94.01,
"learning_rate": 2.7337132953697554e-08,
"loss": 0.7847,
"step": 198000
},
{
"epoch": 95.0,
"learning_rate": 6.834750376549792e-09,
"loss": 0.8535,
"step": 199000
},
{
"epoch": 95.01,
"learning_rate": 0.0,
"loss": 0.7947,
"step": 200000
},
{
"epoch": 95.01,
"eval_loss": 0.5928256511688232,
"eval_runtime": 10.5671,
"eval_samples_per_second": 473.166,
"eval_steps_per_second": 3.785,
"step": 200000
}
],
"max_steps": 200000,
"num_train_epochs": 9223372036854775807,
"total_flos": 3.3694797993816883e+18,
"trial_name": null,
"trial_params": null
}